Servers connecting to personal knowledge bases, flashcard apps, building/querying knowledge graphs, or providing persistent memory for LLMs.
-
Muralnick/sugo: Local, offline MCP server that converts PDFs, scans, screenshots and Office files into compact Markdown for an agent, with built-in OCR. Read-only tools:
convert_document,estimate_savings(2). Reads only the folders passed with--root; no network access at conversion time. Install:uvx sugo-cli mcp --root <folder>. Transport: stdio; no auth. Python, Linux, macOS, Windows. AGPL-3.0. Official MCP Registry id:io.github.Muralnick/sugo. -
Caprock: Read-only MCP server of Caprock, a local desktop app that records every coding-agent session on your computer. Lets Claude Code, Codex or any MCP client search Second Brain memories and the decisions in force for a repository, find what agents wrote across past sessions, read recent sessions with branches, files and pull requests, and see spend and tokens per day and model, GitHub Actions status and loop alerts. No tool can type into a session, start or stop one, or change a setting; credential-shaped text is redacted. Caprock must be running. Install:
claude mcp add caprock -- caprock mcp(Claude Code) orcodex mcp add caprock -- caprock mcp(Codex). Transport: stdio; auth: none; runs locally, reads Caprock over 127.0.0.1. Go, macOS, Windows, Linux. Apache-2.0. Site: https://caprock.dev. Docs: https://github.com/dspv/caprock/blob/master/docs/mcp.md. -
jakubro/folio: Serves a folder of markdown and data files to the browser as live pages, with read-only MCP tools over the folder: documents, SQL over the folder's data files, health (broken links, missing includes, failing queries, broken rules), what changed, history and digests. Tool calls need a running Folio serving the folder. Install:
npx @jakubro/folio <folder>; MCP server:npx --yes @jakubro/folio mcp. Transport: stdio. GPL-3.0-only. Docs: https://jakubro.github.io/folio/. Official MCP Registry id:io.github.jakubro/folio. -
michielinksee/linksee-memory: Local MCP memory server: one SQLite file on your machine that Claude Code, Cursor, Windsurf, OpenAI Codex and Gemini CLI all read and write, for handing a project over with the reasons attached. Drift detection shows which recorded decisions the code has since left. No account or API key. Install:
npx -y linksee-memory. Transport:stdio; no auth; MIT. -
Cloud Alter Ego: Remote MCP connector that gives every AI chat the same rules, profile, contexts, journals and playbooks, stored as plain Markdown files in the user's own OneDrive, Google Drive, Dropbox, GitHub or Azure DevOps; the server does not keep the content. Tools:
start,open_context,log,remember,new_context,list_files,read_file,write_file,choose_repo(9). Install: remote endpointhttps://cloudalterego.cafeboon.com/mcp, streamable-http, OAuth. Open source template: https://github.com/SebastiaanBoon/cloud-alter-ego -
Mnemosyne-OS/mcp: Memory for coding agents from the Mnemosyne OS desktop app (Windows, macOS, Linux). Folders the user declares become separate vaults, search fuses vectors with a local BM25 channel, and
MNEMO_VAULTSlimits which vaults an agent reaches. Agents can post a status card on the user's canvas. Three tools read Claude Code and Antigravity transcripts to list the other live sessions on the same git worktree, with the app closed. The memory tools need the app running. Install:npx -y -p @mnemosyne_os/mcp mnemosyne-mcp. Transport:stdio; no API key; clients: Claude Code, Claude Desktop, and compatible MCP clients; MIT. Site: https://mnemosyne-os.io. Official Registry:io.github.Mnemosyne-OS/mcp. -
candidloom-hub/sourcebolt: Unofficial MCP server for Google NotebookLM: list and manage notebooks, add sources, get answers with citations, run research and keep notes. One static Go binary for Linux, macOS and Windows, or Docker; cookies stored AES-256-GCM encrypted with the key in the OS keyring, no telemetry. Install:
irm https://raw.githubusercontent.com/candidloom-hub/sourcebolt/main/install.ps1 | iex(Windows),curl -fsSL https://raw.githubusercontent.com/candidloom-hub/sourcebolt/main/install.sh | sh(macOS/Linux), ornpx -y sourcebolt serve. Docker:ghcr.io/candidloom-hub/sourcebolt:0.3.1. Transport:stdio(optional loopbackstreamable-httpwith bearer token); auth: your own Google session, no API key. Clients: Claude Code, Claude Desktop, Cursor, VS Code, Codex, Windsurf, Gemini CLI. Apache-2.0. Official Registry:io.github.candidloom-hub/sourcebolt. -
Talkfirst: Speaking practice for adults learning English, Polish, Lithuanian or Czech (beta): pronunciation scoring, voice conversations with an AI partner and quizzes in MCP Apps windows, spaced-repetition flashcards and audio lessons. Remote streamable-http at
https://mcp.talkfirst.io/mcp, OAuth 2.1 with dynamic client registration, no API key needed; free plan. Registry idio.talkfirst/talkfirst. Docs: https://app.talkfirst.io/docs?lang=en. -
Takibi Base: Official Takibi Base server. Ask your Takibi Base collections anything and get answers with cited passages and a support score. Remote streamable-http at
https://app.takibibase.com/mcp, OAuth 2.1 with PKCE (client ID metadata documents), no API key needed. Registry idcom.takibibase/takibi. Docs: https://takibibase.com/docs/agents. -
Hebby Memories: Hosted long-term memory for AI agents. Stores decisions, setups, project facts and pitfalls as typed memories in a graph that is strengthened by Hebbian learning, brings matching memories into context before the agent asks and consolidates in the background; one memory is shared by all of a user's agents. Remote streamable-http at
https://mcp.app.hebby-memories.com/mcp(same address for all customers), OAuth 2.1 with dynamic client registration and client ID metadata documents. Clients: Claude, Claude Code, ChatGPT, Codex and compatible MCP clients. Paid plans, hosted in Germany. Docs: https://www.hebby-memories.com/en/connector/. Official Registry:com.hebby-memories/hebby. -
kburrus64-max/anansi-haven: Free home base for AI agents with passport, persistent memory, encrypted private storage, a job board, agent directory, and skills library over MCP, A2A, and HTTP. Endpoint:
https://anansi-haven.anansidata.workers.dev/mcp. Transport:streamable-http; free utility tools need no key;register_agentfor memory and house tools. Clients: Claude Code, Cursor, VS Code, Windsurf, Codex, Gemini CLI, Claude Desktop, and compatible MCP clients. License: MIT. Docs: https://github.com/kburrus64-max/anansi-haven/blob/main/docs/SETUP.md. Official Registry:io.github.kburrus64-max/anansi-haven. -
marstudio360/vibedna-journal-mcp: Lets an agent keep one markdown journal per project, read it at session start, append decisions, fixes and handoff notes as work ships, and run ranked full-text search across every journal. Install: clone the repo,
pip install -r requirements.txt, runpython server.pywithJOURNAL_ROOTSset to your projects folder. Transport: stdio. Auth: none. License: MIT. -
claimidx/claimidx: Prior-art failure index for agents — a signed claim index of failures other agents have already paid to solve. Install:
pip install -U "claimidx[server]>=0.7.22"thenclaimidx init --agent NAMEthenclaimidx claim --yes --share-yes. MCP:claimidx-mcp. Transport:stdio; no auth; clients: Claude Desktop, Cursor, Claude Code, and compatible MCP clients; site: https://claimidx.com. Official Registry:io.github.claimidx/claimidx. Apache-2.0. -
ChrystianSchutz/ThreadShelf: Local-first MCP server for inspecting indexed collections, running semantic or exact searches, and retrieving complete conversations from a private archive of ChatGPT, Claude, Google AI Studio, OpenRouter, LM Studio, and Grok histories. Clone and install from source with Node.js 20.19+, index your exports first, then launch the stdio server from the repository root with
npm run mcp --silent. No auth; five tools; MIT. MCP setup. -
reacherwu/continuum: Continuous temporal memory engine for AI coding agents (Cursor, Claude Code, Antigravity) that slashes LLM tokens by 96% across 1,000+ turns with <100µs recall. Strictly bounded at 750 physical slots (<75 KB RAM) in 100% pure Rust standard library with zero external dependencies. Install:
curl -fsSL https://raw.githubusercontent.com/reacherwu/continuum/main/install.sh | bash. Transport:stdio; no auth; clients: Cursor, Claude Code, Antigravity, and compatible MCP clients; AGPL-3.0. Site: https://reacherwu.github.io/continuum/. Official Registry:io.github.reacherwu/continuum. -
JinyangWang27/people-context: Local-first memory for AI agents about the people in your life, with explainable identity resolution, relationships, roles, facts, interactions, reminders, and communication guidance in one user-owned SQLite file. Imports are review-gated, sensitive disclosure is operator-gated, and ordinary operation makes no network calls. Install:
uvx --from people-context people-context. Transport:stdio; no auth. Clients: Claude Desktop, Claude Code, Codex, Cursor, Windsurf, VS Code, and compatible MCP clients. MIT. Official Registry:io.github.JinyangWang27/people-context. -
renezander030/graphiti-local: Temporal knowledge graph with six read-only MCP tools and a separate human-approved CLI write queue. Local inference with Ollama and embedded LadybugDB. Install from source with
uv sync --frozen; prepare the database before runninguv run --frozen graphiti-local. Transport:stdio(no auth) orstreamable-http(bearer token). Tested client: Python MCP SDK. Apache-2.0. Docs: https://github.com/renezander030/graphiti-local/blob/main/docs/local-quickstart.md. -
Prem5123/BookMCP: Local Rust MCP server for searching text-based PDF books and retrieving passages with page citations to inform agent system design and implementation. Ingest through the CLI; MCP access is read-only. Install: native binaries or MCPB bundles from GitHub Releases; run
bookmcp serve. Transport:stdio; no auth; MIT OR Apache-2.0. -
vbcherepanov/total-agent-memory: Local-first persistent memory for coding agents — temporal knowledge graph with fact invalidation, procedural memory, episodic recall, AST codebase ingest, and cross-project analogy across 74 tools. The default hot path makes zero LLM and zero network calls, and every published benchmark is reproducible from artifacts in the repo (LongMemEval R@5 95.1%, LoCoMo 0.607, BEAM 1M 0.448). Install:
pip install total-agent-memory,npx total-agent-memory,brew install vbcherepanov/tap/total-memory, ordocker run ghcr.io/vbcherepanov/total-agent-memory:13.0.3. Transport:stdio; no auth; clients: Claude Code, Codex CLI, Cursor, Cline, Continue, Aider, Windsurf, Gemini CLI, OpenCode; MIT. Site: https://totalmemory.dev. Official Registry:io.github.vbcherepanov/total-agent-memory. -
vshulcz/deja-vu: Reads the session transcripts 25 coding agents already write on this machine and serves any of them work from the others, including sessions from before it was installed. Nothing is written to a memory store: the index is built from the transcripts on disk, and retrieval is lexical — no LLM, no embeddings. Install:
brew install deja-vu,scoop install deja-vu, ornpx @vshulcz/deja-vu. Transport:stdio; no auth; clients: Claude Code, Codex, Cursor, Copilot CLI, OpenCode, OpenClaw, Gemini CLI, Zed, and compatible MCP clients; MIT. Docs: https://vshulcz.github.io/deja-vu/. Official Registry:io.github.vshulcz/deja-vu. -
SmartRoot7/clideck-mcp: Free, open-source remote MCP server for verified, version-aware network knowledge, with deterministic answers, change review, snapshot analysis, upgrade guidance, and topology analysis. Endpoint:
https://mcp.clideck.com/mcp. Transport:streamable-http; no auth. Official Registry:com.clideck/network-knowledge. -
n3wth/r3: Local-first persistent memory MCP server for Claude Desktop, Claude Code, Cursor, and compatible clients. Starts an embedded Redis-backed stdio server with no account or API key; optional cloud sync can be configured separately. Site: https://r3.n3wth.com. Install:
npx @n3wth/r3. MIT. -
Liyuan1992/rawmem: Local-first, append-only evidence ledger for AI-agent sessions. Its bounded read-only MCP server exposes verified status, recent-event summaries, and sealed-archive metadata without write tools or automatic promotion into long-term memory. Install:
python -m pip install "rawmem[mcp] @ https://github.com/Liyuan1992/rawmem/releases/download/v0.7.1/rawmem-0.7.1-py3-none-any.whl"; runrawmem-mcp --ledger /path/to/events.jsonl --scopes read:summary. MCPB is also available from the release. Transport:stdio; no auth; clients: Codex, Claude Code, DeepSeek Harness, and compatible MCP clients; MIT. Docs: https://github.com/Liyuan1992/rawmem#readme. Official Registry:io.github.Liyuan1992/rawmem. -
Liyuan1992/memdsl: Review-gated declarative long-term memory for AI agents. Approved source declarations remain the authority, while its MCP server provides bounded catalog, query, trace, explanation, and lint access with read-only scopes by default. Install with
pip install "memdsl[mcp]==0.9.2"and runmemdsl-mcp --workspace /path/to/workspace --scopes read:summary,read:search; MCPB is also available from the release. Transport:stdio; no auth; clients: Codex, Claude Code, DeepSeek Harness, and compatible MCP clients; MIT. Official Registry:io.github.Liyuan1992/memdsl. -
savestatedev/savestate: Encrypted portable memory for AI agents. Snapshots that survive the chat for Cursor, Claude Code, and Clawdbot. Site: https://savestate.dev/when-the-thread-dies. Install:
npx @savestate/cli. -
lalithbuilds/episoda-alpha-mcp: Sovereign local-first cognitive memory engine with Apple Silicon AMX vector acceleration (1.2M+ vecs/sec), SQLite WAL, 4-Way RRF hybrid retrieval (Vector + Trigram + Graph + ACT-R Decay), and native Obsidian vault sync. Zero cloud egress. Install from source:
uvx --from git+https://github.com/lalithbuilds/episoda-alpha-mcp.git episoda-alpha-mcp. Transport:stdio; MIT. -
datamcp Agent Memory: Hosted remote MCP memory server for structured shared project context. Agents can search prior project work with PostgreSQL full-text search, append structured progress, decisions, bugs, and handoffs, and read reviewed canonical Rules and Project Summary through project-scoped links with controlled access modes. Connect through a generated per-link HTTPS endpoint. Transport:
streamable-http. Auth: bearer. Clients: Claude, Cursor, Codex, and compatible remote-MCP clients. Docs: https://datamcp.app/docs#agent-memory. Official Registry:io.github.mironovisa/datamcp. -
Rafter: Hosted remote MCP server for shared team memory, skills, and agents — Claude, Cursor, and other MCP clients start every session with the team's context. Full-text + entity search, fetch with graph edges (cites/cited_by/links), add_memory, and versioned skill/agent artifacts with workspace and role management. Endpoint:
https://app.heyrafter.xyz/api/mcp. Transport:streamable-http; auth: OAuth 2.1 (browser sign-in, dynamic client registration). Clients: Claude Code, Claude Desktop, Cursor, ChatGPT, and compatible remote-MCP clients. Docs: https://github.com/0xksure/rafter-mcp. -
AbuzzHive: Public Q&A boards where AI agents help each other. Agents look up errors other agents already solved (
find_by_errormatches the exact error while ignoring paths, line numbers and ids), search solved problems with the accepted fix, post a problem when stuck, and solve, confirm or vote on others' problems. Endpoint:https://www.abuzzhive.com/mcp. Transport:streamable-http; auth: OAuth 2.1 (dynamic client registration, approved automatically: no sign-up) or API key. Clients: Claude Code (plugin), Claude Desktop, Cursor, VS Code, Codex, Gemini CLI, and compatible remote-MCP clients. Docs: https://github.com/kubec/abuzzhive.com. Official Registry:com.abuzzhive/abuzzhive. -
devchad-cmd/skilldb-sdk: Search reusable AI agent skills; full content requires a Pro or Studio account. Install:
npx -p skilldb skilldb-mcp. Transport:stdio, orstreamable-httpathttps://skilldb.dev/api/mcpwith OAuth or an API key. The separate anonymoushttps://skilldb.dev/api/mcp/catalogendpoint returns search results and short previews only. MIT SDK. Docs. -
Bevel-Software/Hexis: Git-backed platform for centrally managing, reviewing, and sharing skills, tools, and context with AI agents over MCP. Install:
docker compose up -d. Transport:streamable-http; auth: OAuth 2.1 or bearer key; clients: Claude, ChatGPT, Cursor, and other MCP clients; Apache-2.0. -
Lians-ai/Lians: Local-first bitemporal memory for AI agents with deterministic supersession, point-in-time recall, erasure proofs, and tamper-evident audit history. Install and run with
uvx --from "lians-sdk[mcp]" lians-mcp. Transport:stdio; no API key required for local SQLite mode; Apache-2.0. -
hjqcan/GoodMemory: Local-first, auditable memory layer for Codex, Claude Code, AI apps, and any MCP client. Durable SQLite and embedding-free recall work out of the box; memories remain inspectable, correctable, exportable, and deletable, with writeback opt-in. Install:
npm install -g [email protected]; rungoodmemory setupfor Codex/Claude Code orgoodmemory-mcp --standalone --user-id YOUR_USER_IDfor other MCP clients. Transport:stdio; no local auth; MIT. -
JanYork/llm-wiki-cli: Source-grounded project memory for AI agents with citations, provenance, SQLite/FTS5 retrieval, and optional document and code graphs. Its bounded read-only MCP server exposes one exploration tool. Install:
npm install --global @i-xor/lwc; runlwc serve --mcp. Transport:stdio; no auth; Apache-2.0. -
richardreeze/mrb-api: Remote Streamable HTTP MCP server for verified book recommendations, reading lists, series reading orders, consensus best-of lists, and summaries with original sources. Endpoint:
https://mostrecommendedbooks.com/api/mcp. Transport:streamable-http; no auth. -
Goldentrii/AgentRecall-MCP: Correction-first persistent memory for AI agents. Logs every correction as a
CorrectionRecordwith severity and precision KPI (heeded / retrieved), compounds across sessions via 5 memory layers (episodic, semantic, procedural, narrative, correction). FSRS-lite decay + Modern Hopfield retrieval + RRF fusion. Local markdown (~/.agent-recall/), no cloud, no API keys. MIT. Install:npx agent-recall-mcp. -
DanceNitra/inspeximus: Zero-dependency memory layer and MCP server for AI agents (formerly
mnemo). Governance-first: proof-of-deletion certificates, per-tenant isolation, tamper-evident hash-chained receipts, bitemporalas_of()/history(), an authorized-revert channel, an echo-guard so a restated stale value cannot resurrect a corrected fact, and influence gating that decides what a memory may DRIVE by provenance rather than content. Install:uvx --from "inspeximus[mcp]" inspeximus-mcp(one-shot, no prior install), orpip install "inspeximus[mcp]"then runinspeximus-mcp. Transport:stdio; MIT. -
archetypal-ai/archetypal-ai: Persistent memory MCP server for AI agents with recall, remember, and checkpoint tools. Install:
npx archetypal-ai. -
Worktable: Workspace for people and agents to create and revise docs, interactive HTML, drawings, records, and threads. Endpoint:
https://app.worktable.cloud/api/mcp. Transport:streamable-http; auth: OAuth 2.1. Docs: https://docs.worktable.dev. -
ZengLiangYi/ChatCrystal: Local-first AI PKM memory server for coding conversations. Imports Claude Code, Cursor, Codex CLI, Trae, and GitHub Copilot chats into notes, semantic search, tag graphs, Markdown exports, and MCP memory tools. Install:
npx -y chatcrystal mcp. Transport:stdio; local default needs no auth, cloud/private instances can useCHATCRYSTAL_BASE_URLand optional token.
- GetZikra/zikra : Shared persistent memory for Claude Code, Gemini & Codex. One PostgreSQL-backed pool stores decisions, prompts, requirements, and session logs across sessions, tools, and machines. Hybrid vector + keyword search, auto-save via Stop/PreCompact hooks, half-life confidence decay, role-based access (owner/R&Der/viewer), MIT licensed, self-hosted. (zikra.dev)
-
maxbaluev/accreted-intelligence: Local-first Work Model memory for coding agents. Exposes an MCP stdio server with
acc_retrieveandacc_actso Claude Code, Codex, OpenCode, Cursor, and other clients can retrieve scored memory, record actions, and learn from real outcomes. Install viacurl -fsSL https://raw.githubusercontent.com/maxbaluev/accreted-intelligence/main/bootstrap/install | sh. -
thegobi/ithz_mcp: Local-first deterministic project memory for AI coding agents. Provides project-owned context packs, decisions, gates, risks, scoped claims, reviewer notes, and explicit checkpoints while Git remains the code transport. Install:
pip install ithz-mcp. Transport:stdio; no cloud auth required for local use. -
screenpipe/screenpipe: Local-first workflow memory for AI agents. It captures selected screen, audio, app, and meeting context locally so Claude, Codex, Cursor, and local AI workflows can search real work history through MCP/API, then turn daily work into cited notes, SOPs, workflow reports, and automation candidates while keeping sensitive context under user control. Install:
npx -y screenpipe-mcp@latestwith a running local recorder andSCREENPIPE_LOCAL_API_KEY(desktop Settings > Connections configures this); docs: https://github.com/screenpipe/screenpipe/blob/main/packages/screenpipe-mcp/README.md. Source-available under the Screenpipe Commercial License /LICENSE.md. -
a2cr/a2cr: MCP server for AI-agent handoffs. Saves client-encrypted WorkBaton checkpoints and WorkStash notes so Codex, Claude Code, Roo Code, and other MCP clients can resume work without passing full chat history.
-
7xuanlu/wenlan: Local-first AI knowledge base and LLM wiki. Distills agent work into source-cited pages with session handoffs, graph context, and hybrid retrieval across MCP clients.
-
Beever-AI/beever-atlas: Open-source LLM Wiki that distills team-chat corpora (Slack, Discord, Microsoft Teams, Mattermost, Telegram) into a 3-tier semantic schema and an entity graph. Exposes 16 MCP tools for cited Q&A from Claude Code and Cursor. Apache 2.0, runnable in 3 commands via
make demo. -
STiFLeR7/memex: Developer context continuity system. Watches your git repos and builds a temporal knowledge graph of modules, symbols, decisions, and open problems via Graphiti + Neo4j, then serves it to any AI coding agent over MCP. Every edge carries a validity window and a confidence score that decays over time. 12 tools across read and write. Install via
npx -y stifler-memex-mcp. MIT licensed. -
novyxlabs/novyx-mcp: Persistent memory for AI agents with 107 tools: remember, recall, rollback, audit, knowledge graph, governed actions, eval, cortex, and Runtime v2 agent orchestration. Local SQLite (zero-config) or Novyx Cloud. Install via
uvx novyx-mcp. MIT licensed. -
yantrikos/yantrikdb-mcp: Cognitive memory for AI agents — persistent semantic memory with knowledge graph, adaptive recall, conflict detection, entity relationships, and memory consolidation across sessions. 24 tools. Works with Claude Code, Cursor, Windsurf, and any MCP client.
-
nicholasglazer/gnosis-mcp: Zero-config MCP server that loads markdown docs into SQLite (FTS5 keyword search) or PostgreSQL (pgvector hybrid semantic+keyword search) and exposes 6 tools and 3 resources for AI agents to search, retrieve, and manage documentation.
-
varun369/SuperLocalMemoryV2: A local-first AI memory system providing persistent memory across 17+ AI tools via MCP, featuring knowledge graphs, hybrid search, and pattern learning with zero cloud dependency.
-
omega-memory/core: Persistent memory for AI coding agents with semantic search, auto-capture, intelligent forgetting, and cross-session learning. #1 on LongMemEval benchmark (95.4%). 83 MCP tools (26 core + 57 pro). Multi-agent coordination, knowledge graphs, ONNX embeddings. Local-first with zero cloud dependency.
-
gdcc/mcp-dataverse: Facilitates multilingual data integration and exploration in Dataverse using Croissant ML.
-
singularityjason/lightning-memory: Persistent agent memory for the Lightning economy. Vendor reputation tracking, spending anomaly detection, Nostr identity (BIP-340), and L402 payment gateway for agent-to-agent knowledge markets. Local-first SQLite with FTS5 search.
-
xsp52Hz/cognigraph-mcp-server: CogniGraph MCP Server generates mind maps, relationship graphs, and knowledge graphs using CLI tools and AI analysis, compatible with various local MCP clients.
-
johnhenry/mcp-server-ipfs-context: Facilitates storing and retrieving conversational context using IPFS for seamless session continuity.
-
upamune/duckdb-hybrid-doc-search: Facilitates hybrid indexing and search of Markdown documents using DuckDB, integrating full-text and vector search capabilities with AI coding agents via an MCP stdio server.
-
beihaili/Get-Started-with-Web3: Read-only MCP server exposing a bilingual Web3 curriculum with 58 lessons, glossary search, learning paths, llms.txt, and citation-ready context packs for AI tutors and coding agents.
-
ipospelov/mcp-memory-bank: Facilitates structured documentation for context preservation in AI environments through a Memory Bank system.
-
brian3814/notion_fastmcp: Facilitates task management in Notion through an MCP server integration with Cursor IDE.
-
KerolIA98/mcp-sample: A template MCP server integrated with Mem0, enabling AI agents to manage long-term memory through semantic search and storage.
-
mkusaka/mcp-server-memory: A persistent memory server utilizing a local knowledge graph to enable Claude to retain user information across interactions.
-
jordankamto/code-explorer-mcp: A TypeScript-based server that manages and summarizes text notes using the Model Context Protocol.
-
meetdhanani17/xgmem: A TypeScript-based MCP server for managing project-specific and cross-project knowledge graph memory for LLM agents and tools.
-
Svtter/chatdb: ChatDB serves as a memory layer for GPT by recording conversations in SQLite, enhancing conversational context retention.
-
tejasPhaveri/custom-instructions: Enhances the ultimatememory system with Context7 and Sequential Thinking MCP tools for improved memory management and task execution.
-
Im-neko/mcp-wikijs: Facilitates AI model interaction with WikiJS content through a TypeScript-based MCP server, enabling document search, creation, and management.
-
fazer-ai/mcp-obsidian: Facilitates interaction between Claude and Obsidian vaults via a Local REST API, enabling seamless note management and search capabilities.
-
fmicalizzi/mcp-read-json: Facilitates reading and querying JSON-based knowledge bases for seamless integration with LLMs like Claude.
-
PyneSys/project-mem-mcp: Facilitates AI agents in maintaining persistent project memory through a memory file system, enabling seamless information retrieval and updates during interactions.
-
ndlxp2008/mcp-temple: A TypeScript-based MCP server implementing a simple note-taking system with URI-based resource management and note summarization tools.
-
dwarvesf/mcp-playbook: Facilitates project documentation management and conversation log saving within a specified directory using Node.js and GitHub API interactions.
-
KawaroX/codex-vitea-mcp: A specialized MCP server for ViteaOS that connects AI assistants with MongoDB to manage personal information, including item location, travel time estimation, contact management, biodata analysis, and task tracking.
-
jzumwalt/git-mcp: GitMCP transforms GitHub projects into documentation hubs, enabling AI tools to access up-to-date documentation and code, reducing hallucinations and improving code accuracy.
-
shifusen329/doc-lib-mcp: Facilitates document ingestion, chunking, semantic search, and note management with a custom URI scheme and robust content extraction tools.
-
kiranraathod/taskflow-memory-server: A task management server with persistent memory architecture, integrating Claude AI for intelligent planning and task execution using the Model Context Protocol.
-
kaz56-t/mcp-latest-document: A Model Context Protocol server offering access to the latest documentation for various services.
-
angrysky56/NeoCoder-neo4j-ai-workflow: Enables AI assistants to utilize a Neo4j knowledge graph for dynamic coding workflows and project memory management.
-
lyonzin/knowledge-rag: Local RAG system for Claude Code via MCP with hybrid semantic + BM25 search, cross-encoder reranking, and 12 MCP tools — no external servers required.
-
CaptainCrouton89/alaria-wiki-mcp: A boilerplate for creating an MCP server that stores and retrieves information using vector embeddings, enabling semantic search and integration with AI assistants like Claude.
-
CaptainCrouton89/mcp-boilerplate: A boilerplate for creating an MCP server that utilizes vector embeddings for storing and retrieving information, enabling semantic search and integration with AI assistants like Claude.
-
olafgeibig/knowledge-mcp: Facilitates AI agents in accessing specialized knowledge bases through a Model Context Protocol server interface, enhancing their domain-specific reasoning capabilities.
-
tonytimo/onboarding-mcp: Orellis enables engineers to query and explore their local Python codebase using natural language, ensuring secure and private onboarding assistance through a local LLM.
-
lbdudc/mcp-fm-analyzer: Analyzes feature models in UVL, offering tools for insights like atomic sets and branching factors.
-
swax/OneNoteMCP: Facilitates seamless access to personal OneNote notebooks via an updated MCP API with enhanced authentication and performance features.
-
cdgaete/token-scope-mcp: TokenScope offers token-aware directory exploration and analysis for LLMs, enhancing codebase understanding through intelligent summaries and comprehensive reporting.
-
simon-k/knowledge-mcp: Facilitates knowledge retrieval from a Neo4J database through an MCP server interface.
-
Disdjj/mmp-mcp: Facilitates AI model memory management through a structured tree protocol, enabling efficient storage and retrieval of interaction-generated memories.
-
lethain/library-mcp: Facilitates interaction with Markdown knowledge bases by retrieving and managing content through tags, text, slugs, URLs, and date ranges.
-
Going34/memory-bank-mcp-npm: Facilitates AI assistants by managing and updating memory banks with project code changes and user prompts.
-
daheepk/arxiv-paper-mcp: Facilitates efficient search, exploration, and comparison of arXiv papers using a custom MCP framework.
-
jin38324/oci-documentation-mcp-server: Facilitates searching and converting OCI documentation into markdown format for streamlined access and integration.
-
HaruHunab1320/docmost-mcp: Docmost integrates AI assistants with collaborative documentation through the Model Context Protocol, enabling seamless AI-assisted workflows.
-
gpreddy172/MCP-Server-for-Documentation: Facilitates searching for the latest documentation across various libraries using a standardized protocol.
-
SilenceBoy/ai2flomo: Facilitates seamless note-taking and management between AI assistants and Flomo, leveraging FastMCP for efficient synchronization and historical record retrieval.
-
AdilFayyaz/Slack-Search: A local MCP server that enhances Slack search capabilities by integrating with Claude Desktop for intelligent, private querying of Slack export logs.
-
jfim/obsidian-tasks-mcp: Facilitates AI-assisted task management by extracting and querying tasks from Obsidian markdown files using the MCP protocol.
-
andrewginns/mcp-mind-palace: A Python-based MCP server offering a persistent, searchable knowledge base for LLMs using Markdown files and ChromaDB for semantic search.
-
sho0318/ndl-mcp: Facilitates AI-driven book searches and data retrieval using the Next Generation Digital Library API.
-
gtfierro/rdf-mcp: Facilitates the use of Brick and 223P ontologies through Model Context Protocol servers, enabling ontology-based operations and definitions retrieval.
-
fagnercarvalho/ontario-fish-stocking-mcp: Fetches and queries fish stocking data from Ontario, Canada using various tools like coordinates, species, and year.
-
kiseki-technologies/kiseki-labs-readwise-mcp: Facilitates programmatic access and manipulation of Readwise documents and highlights via the Readwise API.
-
harukikaneko/doc-qa: A document question-answering system leveraging DuckDB-VSS and Python for efficient document search using Mean Contextualized Pooling.
-
aixiasang/weread-mcp: Facilitates AI agents to interact with and manage your WeRead library, enhancing reading content analysis.
-
shariqriazz/llms-txt-mcp: A comprehensive MCP server for managing and searching local RAG documentation systems with web discovery and AI-powered tools.
-
yangzec/cubox-mcp: A TypeScript-based server implementing a notes system with tools for note creation and summarization.
-
zw459123678/tuniao-server: Access and explore TuNiao UI components through a Model Context Protocol server with AI model integration.
-
Joeycz/whistle-docs-server: Facilitates search and access to Whistle documentation through a TypeScript-based MCP server.
-
StitchAI/stitch-ai-mcp: A decentralized knowledge hub for AI, enabling efficient memory management through a Model Context Protocol server.
-
kibela/kibela-mcp-server: Facilitates seamless interaction with Kibela through a local MCP server, enabling note and folder management, AI-assisted writing, and more.
-
noahgorstein/mcp-server-stardog: Facilitates interaction and automation with Stardog Knowledge Graphs through a Model Context Protocol server.
-
springdo/mcp-cookbook: Access the Gousto cookbook API to browse and search recipes with this MCP server.
-
thestumonkey/notion-chakra-mcp: Facilitates interaction with Notion databases and pages through customizable MCP endpoints, supporting both stdio and SSE transports.
-
enjoyzl/mcp-server-confluence-ts: Facilitates seamless interaction with Confluence through a robust MCP-based API service, offering features like space information retrieval, content search, and detailed page access.
-
casheiro/synapstor: Synapstor offers a comprehensive solution for semantic storage and retrieval of information using vector embeddings and integrates seamlessly with LLMs through the MCP protocol.
-
aiuluna/knowledge-graph-mcp: Facilitates the creation, management, and visualization of knowledge graphs with seamless integration into MCP-compatible AI assistants.
-
cuongpham2107/word-mcp-server: Facilitates the creation and editing of Microsoft Word documents via API, leveraging FastMCP for seamless LLM integration.
-
dilboy/markitdown-mcp: Facilitates the conversion of various URI types to Markdown using a lightweight MCP server with STDIO and SSE support.
-
Dr-Aniekan-Udo/Notetaker-MCP-Server: Facilitates seamless note-taking and management through Claude desktop integration, utilizing MCP server capabilities.
-
sosacrazy126/greptile-mcp: Integrates with the Greptile API to enable AI agents to perform natural language code search and querying across repositories.
-
okooo5km/memory-mcp-server-go: Facilitates knowledge graph management for LLMs, enabling persistent memory across conversations.
-
lkm1developer/google-docs-mcp-server: Facilitates seamless integration with Google Docs API, allowing AI assistants to manage documents through a robust MCP server.
-
aliyun/mcp-server-esa: Facilitates communication between AI models and Edge Security Acceleration services using a standardized protocol.
-
keepondream/wordnet-mcp-server: A WordNet-based MCP server offering vocabulary query functions such as synonyms, antonyms, hypernyms, hyponyms, and definitions.
-
Adityaakr/vara-mcp-updated: Facilitates AI applications' interaction with the Vara Network by providing onchain tools for retrieving block and transaction details, checking account balances, and tracking transaction history.
-
kbyk004/my-docs-mcp-server: Indexes and searches Markdown files in a directory, enabling full-text search and file retrieval via the Model Context Protocol.
-
ReyNeill/Kontxt: Facilitates codebase indexing by connecting to local repositories and providing AI clients with contextual analysis using Gemini 2.0 Flash.
-
a-bonus/google-docs-mcp: Facilitates seamless interaction between Claude and Google Docs, enabling advanced document manipulation and formatting through MCP.
-
7nohe/confluence-mcp-server: Facilitates interaction with Confluence pages through a Model Context Protocol server using the Confluence API.
-
MartinM85/mcp-server-graph-api: Facilitates AI-driven interactions with Microsoft Graph API using a C# MCP server.
-
julianehuettl/wolfram-alpha-mcp: Integrates Wolfram Alpha's computational knowledge engine with LLMs for advanced mathematical and scientific queries.
-
stephen9412/memos-mcp-server: Integrates Memos API with AI assistants for seamless memo management, including search, creation, and retrieval functionalities.
-
cconstable/obsidian-mcp: Facilitates reading and searching Obsidian vaults through an MCP server, enhancing note accessibility and integration with tools like VS Code.
-
zohar/knesset-mcp: Facilitates AI applications in accessing and interacting with Israeli Knesset parliamentary data through a standardized interface.
-
blck-snwmn/repodocs: Facilitates document management and access within repositories using a simple MCP server setup.
-
zereight/google-docs-mcp: Facilitates access, reading, and editing of Google Docs documents through an MCP server.
-
HappymanOkajima/apm-mcp-server: Facilitates interaction with the Agile Practice Map knowledge base through a large language model.
-
Safe-Swiss-Cloud-AG/mcp_pdf_reader: Facilitates reading PDF documents through an MCP server, compatible with Claude Desktop and LibreChat.
-
Moe03/mcp-hello-world: Facilitates interaction with Google Maps API for geocoding, reverse geocoding, place searches, and more.
-
kfischer-okarin/mcp-server-python-functions: A Python-based MCP server utilizing 'just' for streamlined command execution and integration with MCP clients like Claude.
-
yaojiafeng/qdxs-add-mcp-server: A Model Context Protocol server utilizing the mcp-framework, designed for easy tool integration and deployment.
-
emekaokoye/mcp-rdf-explorer: Facilitates conversational exploration and analysis of RDF-based knowledge graphs using SPARQL queries in local or endpoint modes.
-
coleam00/mcp-mem0: Facilitates AI agents with persistent memory capabilities using Mem0, serving as a template for building MCP servers with Python.
-
tomiyan/mcp-gbiz-info: Facilitates the retrieval of corporate numbers using the gBizINFO API.
-
TenzDelek/monlamTranslation-mcp: Facilitates Tibetan translation through a local MCP server setup.
-
egeokt/mcp-server: Facilitates access to the latest documentation for tools like LangChain, OpenAI, and Lama-Index.
-
onikun94/markdown-to-notion-mcp: Transforms Obsidian markdown into a Notion-compatible format using the Model Context Protocol (MCP) server.
-
shariqriazz/mcp-ragdocs: Facilitates AI assistants in augmenting responses with relevant documentation context through vector search, supporting multiple embedding providers and using Qdrant as the vector database.
-
danosb/onenote-mcp: Facilitates AI language models' interaction with Microsoft OneNote, enabling seamless access, creation, and analysis of notes.
-
grainne-b/notion_mcp_server: Facilitates integration with Notion through an MCP server, enabling seamless API interactions and automation.
-
mjpitz/mcp-rfc: Facilitates programmatic interaction with RFC documents by fetching, parsing, and reading them from ietf.org.
-
esurovtsev/know-flow-mcp: KnowFlow enhances LLMs by dynamically retrieving structured knowledge from external sources via the Model Context Protocol.
-
jimmyliou777/mcp-test: A Node.js server leveraging MCP SDK for custom tool and resource implementation, featuring basic math operations and Puppeteer-based automation.
-
shiwenbin1617/Google-Scholar-MCP-Server: Facilitates AI-driven academic research by enabling seamless access to Google Scholar papers and author information through an MCP interface.
-
loganrooks/zlibrary-mcp: Facilitates AI coding assistants in accessing Z-Library for book searches, metadata retrieval, and download links via a standardized API.
-
bchopson/indy-gov-mcp: Fetches trash pickup day information for Indy.gov using an MCP server.
-
Tomobobo710/MCP-Codebase-Browser: A Python-based MCP server enabling Claude to manage and manipulate codebases on Windows systems.
-
readwiseio/readwise-mcp: Facilitates seamless integration between LLM clients like Claude and Readwise by acting as a local MCP server bridge.
-
flight505/mcp-think-tank: MCP Think Tank enhances Claude's reasoning and memory capabilities through structured thinking and a persistent knowledge graph.
-
egoist/raindrop-mcp: Facilitates interaction with Raindrop.io for managing and searching bookmarks through MCP protocol.
-
docdyhr/simplenote-mcp-server: Integrates Simplenote with Claude Desktop, enabling advanced note management and search capabilities through an MCP server.
-
harukikaneko/esa-mcp: Facilitates integration with ESA through a customizable MCP server setup.
-
Harry-027/JotDown: Facilitates seamless integration of LLMs with Notion for page creation and mdBook generation, enhancing content management workflows.
-
godwin3737/mcp-server-microsoft365-filesearch: Facilitates advanced file search and metadata analysis within Microsoft 365, integrating seamlessly with SharePoint and OneDrive for efficient business workflows.
-
gohluke/dayze-mcp: Pay-per-call people + life-in-days knowledge API for AI agents. Notable packs with day-of-life numbers, similar people, and birthday peers. Transport:
streamable-httpcompatible REST (GET/POST); Auth: none for public tools, x402 USDC on Base after free tier; Endpoint:https://dayze.com/api/v1/mcp; Docs: https://dayze.com/.well-known/mcp.json; License: see repo. -
kyrelldixon/notion-mcp: Facilitates AI assistants' interaction with Notion workspaces by providing a standardized interface for accessing and manipulating Notion content.
-
SiroSuzume/my-mcp-server-template: A template for setting up and configuring MCP servers using Node.js.
-
linxule/lotus-wisdom-mcp: A structured problem-solving tool using the Lotus Sutra's wisdom framework, offering visualizations and multi-perspective approaches for complex issues.
-
kingsarthucodes/MCPLessonGenerate: Integrates with Firecrawl for advanced web scraping, crawling, and content extraction with robust error handling and logging.
-
VmLia/books-mcp-server: Facilitates book querying through MCP clients like Cherry Studio, offering seamless integration and configuration.
-
zamd/remote-mcp-server: Deploy a remote MCP server on Cloudflare Workers with OAuth login and connect it to Claude Desktop for seamless tool integration.
-
yellowgg2/mcp-bookstack: Facilitates searching and retrieving structured data from BookStack pages using customizable queries and clean HTML-to-text conversion.
-
dragonjump/mcp-ARCknowledge: Bridge all your custom webhook endpoints into a unified configuration for efficient knowledge base management and querying.
-
R3Wlad/mcp-notion-server: Facilitates interaction between Claude and Notion workspaces through a dedicated MCP server.
-
sifue/zen-syllabus-mcp: Facilitates access to ZEN University's syllabus content through an MCP server implementation.
-
D-lyw/ckb-docs-mcp: Facilitates semantic search of CKB ecosystem documents and code using the NervePuppy API and vector database integration.
-
SonLee369/mcpdemo: A comprehensive collection of reference implementations for the Model Context Protocol, showcasing secure and controlled access to tools and data sources for LLMs.
-
angelsen/grail: Grail is an AI-powered pattern management system for Claude Code, enabling efficient organization, discovery, and application of code patterns through semantic search and MCP server integration.
-
veainilo/awesome-cursor-mcp.json: Enhance Cursor editor's AI capabilities with pre-configured MCP server setups for browser interaction, sequential thinking, and Playwright automation.
-
buger/docs-mcp: Facilitates AI-driven searches within GitHub repositories or documentation by transforming them into MCP servers, enabling natural language queries.
-
zxkane/quip-mcp-server: Facilitates interaction with Quip spreadsheets by retrieving and converting data into CSV format, with support for authentication and large data handling.
-
BangyiZhang/xmind-generator-mcp: Facilitates the creation of structured Xmind mind maps through LLMs using the MCP protocol.
-
fnf-deepHeading/mcp-api-server-wrapper: A template for wrapping external APIs into MCP servers using AI, featuring automatic API integration and Docker support.
-
apw124/logseq-mcp: Facilitates AI agents' interaction with Logseq graphs through a set of MCP tools, enabling seamless data manipulation and retrieval.
-
rarandeyo/INIAD-MOOCs-MCP: Automates login, assignment submission, and file uploads for INIAD MOOCs using Playwright MCP.
-
moritalous/mermaid-doc-mcp-server: Facilitates the generation and retrieval of Mermaid diagram documentation through an MCP server interface.
-
EnzoVezzaro/mcp-dominican-layer: Access detailed information about the Dominican Congress, including legislative agendas and new bills, through a dedicated MCP server.
-
OpenMined/DistributedKnowledge: Distributed Knowledge is a decentralized, network-aware LLM system that leverages a federated architecture for secure and private data exchange, fully compatible with MCP hosts.
-
cer12u/growi-mcp: Repository classified as category 19
-
Rwb3n/obsidian-mcp: Facilitates interaction with an Obsidian vault by providing tools for note management, metadata handling, and content searching.
-
nirvash/bookmark-mcp: Facilitates Chrome bookmark management through LLM commands using a Machine Control Protocol server.
-
Jacobinwwey/Notemd-mcp: A TypeScript-based MCP server enabling a notes system with multi-client AI support, including Claude and Cline integrations.
-
vyahhi/hyperskill-mcp-server: Facilitates searching and accessing educational content on Hyperskill through an MCP server.
-
7nohe/qiita-mcp: Facilitates interaction with Qiita's API for retrieving user and article information using the Model Context Protocol.
-
r-huijts/strava-mcp: Connects to Strava API, enabling LLMs to access and manage Strava data and activities.
-
PhiloSolares/roam-mcp: Connects AI assistants like Claude to Roam Research for seamless data interaction and management.
-
jaw9c/awesome-remote-mcp-servers: A curated list of high-quality remote MCP servers for developers seeking reliable and production-ready services for AI applications.
-
mbcrawfo/KnowledgeBaseServer: Facilitates LLMs in storing and retrieving conversational memories using SQLite's full-text search capabilities.
-
ruliana/mcp-pkm-logseq: Facilitates interaction with Logseq's Personal Knowledge Management system through customizable instructions.
-
FrankGoortani/cv-mcp: A specialized MCP server offering structured access to Frank Goortani's CV data for AI assistants and MCP-compatible clients.
-
kw510/strava-mcp: Integrates Strava OAuth with MCP clients using Cloudflare Workers for secure authentication and tool access.
-
Xiawpohr/mcpilot: MCPilot empowers LLMs to securely interact with blockchains via MetaMask, ensuring private keys remain protected while enabling seamless Web3 experiences through natural language commands.
-
SiamHas/MCP_TO_PC_ENTITIES: Facilitates querying DNO APIs for catalog entities and pricing rules via Claude integration.
-
Qwinty/anytype-mcp: Facilitates interaction with Anytype data through an MCP server, enabling AI assistants and other clients to manage spaces, objects, and templates.
-
BaoxingZhang/mcp-server-notion-prompt: Access and manage prompts stored in Notion databases using the Model Context Protocol (MCP) standard.
-
phkus/tinderbox-mcp: Facilitates AI-driven interactions with Tinderbox, enabling natural language control and integration with other services through AppleScript.
-
hassaku63/mcp-start-go: A Go-based example of an MCP server implementation, demonstrating cross-platform capabilities with StdioTransport.
-
JordanDalton/AnthropicMcpClient: A CLI tool for interacting with Anthropic MCP servers, enabling tool execution across multiple server connections.
-
jesegher/EduMCPServer: Facilitates AI-driven management of Microsoft Education classes, assignments, and rubrics through integration with Microsoft Graph API.
-
199-bio/better-thinking-mcp: Facilitates structured, multi-step reasoning inspired by LLM computational mechanisms, enhancing transparency and robustness in decision-making processes.
-
rsprudencio/binja_mcp: Facilitates interaction and automation with Binary Ninja using Large Language Models through the Model Context Protocol.
-
199-bio/occams-razor-mcp: Guides LLMs through a structured, simplicity-focused problem-solving process for coding tasks, enhancing code quality and alignment with user intent.
-
urbushey/ts_mcp: A TypeScript-based MCP server designed for integration with Claude Desktop, facilitating seamless tool management and execution.
-
MeirKaD/SBR-MCP: Facilitates browser automation tasks such as navigation, form filling, and data retrieval through an MCP server integration with Claude Desktop.
-
alan-meigs/MCP_EXP: A terminal-based chat client with MCP server integration, featuring OpenAI API connectivity and a weather service using the National Weather Service API.
-
p-funk/FEGIS: FEGIS empowers LLMs with structured, persistent memory through customizable cognitive tools, enabling dynamic schema-defined cognition and semantic retrieval.
-
JackKuo666/Crossref-MCP-Server: Facilitates AI-driven searches and metadata retrieval of academic papers via Crossref using MCP.
-
flyer103/notion-mcp: Facilitates AI model interactions with Notion documents and data through a Model Context Protocol server.
-
closerbrasil/remote-mcp-server: Deploy a remote MCP server on Cloudflare Workers with OAuth login and connect it to Claude Desktop for seamless tool integration.
-
RedSiamese/local-knowledge-mcp: Local Knowledge manages and provides a local knowledge base for dynamic storage, updating, and querying of knowledge, tailored for large language models to access workspace-relevant information.
-
pfldy2850/mcp-openmetadata: Facilitates seamless integration of OpenMetadata APIs with FastMCP for efficient metadata management.
-
shiquda/mediawiki-mcp-server: Facilitates seamless interaction with MediaWiki-based sites, enabling LLMs to search and retrieve content from platforms like Wikipedia and Fandom.
-
Tae4an/mcp-prompt-manager: Facilitates efficient management of local prompt files for AI models like Claude, enabling prompt creation, retrieval, modification, and deletion.
-
uoky5217/Mcp-Server-Study: A learning project implementing an MCP server with stdio and SSE transport methods for basic arithmetic operations.
-
kwanLeeFrmVi/mcp-server-memory: Facilitates AI models' persistent memory through a knowledge graph, enabling information retention and recall across interactions.
-
yenslife/nckuhub-mcp: Leverage the NCKUHub API to provide a robust MCP service for seamless integration and development.
-
kaygee/rev-mcp: Fetches transcripts from Rev.com using their API, integrating seamlessly with MCP clients like Claude for Desktop.
-
dpk83/graphmcpserver: Facilitates seamless integration with Microsoft Graph via REST API using MCP.
-
RohitKrish46/docs-mcp-server: Empowers LLMs to dynamically search and retrieve up-to-date documentation from popular AI libraries using a lightweight MCP server.
-
khanhtoanmeo/notion_mcp: A Node.js-based MCP server that integrates with AI-assisted development environments, offering tools for arithmetic operations and environment variable retrieval.
-
ChenyqThu/mcp-server-weread: A lightweight server bridging WeRead data with Claude Desktop, enabling seamless access to reading notes and data.
-
Lvigentini/LTC-RAG-MCP: Enhances Claude with long-term memory capabilities using a Weaviate-based RAG system for seamless retrieval of past interactions and external knowledge sources.
-
Ssiswent/mcp-server-bear: Facilitates note creation, searching, and management in Bear via conversational interactions with MCP clients.
-
MahithChigurupati/Confluence-MCP-Server: FastMCP-based server enabling seamless programmatic access to Confluence's REST API for managing spaces, pages, and executing content searches.
-
AnkitPatel12/StravaMCP: Connects to Strava's API to enable AI assistants like Claude to access and analyze fitness data.
-
uta8a/linkding-mcp-server: Facilitates integration with the linkding bookmark manager through a customizable MCP server.
-
bradleygolden/hexdocs-mcp-server: A TypeScript server offering semantic search for Hex package documentation, leveraging the Model Context Protocol for seamless AI integration.
-
ronniemh/phrases-MCP-server: An elegant and efficient MCP server for managing inspirational phrases, seamlessly integrating with Claude for Desktop and other MCP clients.
-
yuzhi-jiang/memos-mcp: Connects to Memos for structured access and interaction via AI assistants like Claude, offering features like search, creation, and management of memos.
-
jaohbib/MCP-for-VuFind: Facilitates the integration of VuFind's Swagger API into LLMs for literature search capabilities.
-
ChenyqThu/mcp-server-memos: Facilitates natural language interaction with Memos note service via MCP protocol, supporting note creation, search, and updates with secure token authentication.
-
1nbuc/mcp-integration-suite: Facilitates seamless integration with SAP Integration Suite by managing and deploying integration artifacts like IFlows and Message Mappings.
-
rbownes/mcp-memory-server: A Rust-based MCP server offering memory storage and retrieval with support for multiple storage backends and embedding models.
-
entanglr/zettelkasten-mcp: Facilitates the creation, linking, and searching of atomic notes using the Zettelkasten methodology, integrated with Claude for AI-assisted knowledge management.
-
wolderufael/docs-MCP-server: A Model Context Protocol server implementation for integrating with various LLM clients, offering flexibility and secure data handling.
-
hetaoBackend/mcp-github-trending: Provides a simple API interface to access GitHub's trending repositories and developers data, with filtering options by language and time period.
-
jespino/developers-conferences-agenda-mcp: Facilitates access to developer conferences and events data, allowing users to search, filter, and retrieve information through MCP-compatible clients.
-
r-huijts/oorlogsbronnen-mcp: Access Dutch WWII archives through AI-powered natural language queries using the Oorlogsbronnen MCP server.
-
abeyuya/memory-bank-loader-mcp: Efficiently retrieves and organizes files from a memory-bank directory to streamline AI tool usage and reduce operational costs.
-
valentinludu/oblio-mcp: Facilitates seamless interaction between Claude and Oblio.eu accounting software through a robust MCP server implementation.
-
scotthelm/wikipedia-mcp-server: Facilitates AI-driven interactions with Wikipedia through a Model Context Protocol server, enabling content retrieval, historical event searches, and image access.
-
xianminx/mcp-server-flomo: Facilitates seamless note creation in Flomo through AI chat interactions using natural language commands.
-
laughall-miyake/udemy-instructor-mcp: Node.js server for accessing Udemy Instructor API, enabling retrieval of courses, questions, and reviews.
-
OfWind/graph-memory-mcp: A persistent memory server utilizing a local knowledge graph to enable Claude to retain user information across interactions.
-
hulkworks/readthedocsmcp: Facilitates LLM interaction with Read the Docs documentation through a Multi-Context Protocol server.
-
MushroomFleet/LTM-CLINE: LTM-CLINE enhances Claude 3.7 with long-term memory and persona evolution capabilities using SQLite, integrated with the Cline VSCode extension.
-
gannonh/memento-mcp: Memento MCP offers a scalable knowledge graph memory system with semantic retrieval and temporal awareness for LLMs, leveraging Neo4j for storage and vector search.
-
s2005/mcp-memory-server: Enables persistent memory for Claude using a local knowledge graph to remember user information across chats.
-
kargarisaac/Hekmatica: Hekmatica is a deep research agent that uses BAML and LangGraph to answer questions by leveraging web search, external tools, and LLMs.
-
anivenk25/Zettelkasten: Zettelkasten is an MCP server tailored for chatbot applications, utilizing hybrid data storage with Pinecone and Neo4j for efficient cross-session chat context management.
-
jeremytandjung/my-notion-mcp-server: A fast MCP server leveraging Bun for seamless integration with Claude Desktop and Notion.
-
boetro/obsidian-mcp: Facilitates interaction with Obsidian vaults through LLMs, enabling file management and content updates.
-
Phasoric: Search, read, and safely update Markdown notes in explicitly connected Phasoric knowledge vaults. Connect:
https://phasoric.com/api/mcp. Transport:streamable-http; Auth: OAuth; requires a Phasoric account with MCP access (Premium or higher) and an explicit vault mapping. Proprietary hosted service; local-only vaults are not remotely accessible. Tools:list_vaults,search_notes,get_note_content. -
beshkenadze/mcp-obsidian: Facilitates seamless interaction with Obsidian through its Local REST API, offering versatile server implementations and transport modes.
-
Meru45/Zettelkasten: Zettelkasten is an MCP server tailored for chatbot applications, utilizing hybrid data storage with Pinecone and Neo4j for efficient cross-session chat context retrieval.
-
zhaoyi48/chinese-holidays-MCPServer: A TypeScript-based server for querying Chinese public holidays via the MCP protocol.
-
maoruibin/mcp-server-inbox: Facilitates note creation through any MCP client by integrating with the inBox API, allowing users to create notes in a conversational manner.
-
monotykamary/elevenlabs-docs-mcp: Facilitates structured access and search of ElevenLabs documentation via the Model Context Protocol.
-
obutora/writing-style-mcp: Facilitates writing style adjustments through an MCP server with Go-based execution.
-
chatmcp/mcp-server-flomo: Facilitates seamless note-taking to Flomo through a TypeScript-based MCP server.
-
DataScienceDisciple/obsidian-mcp: A TypeScript-based MCP server enabling AI models to interact with Obsidian vaults via the Local REST API, featuring advanced search and content management tools.
-
esakrissa/mcp-doc: Facilitates integration between LLM applications and documentation sources via the Model Context Protocol, providing access to LangGraph and MCP documentation.
-
wangle201210/dify-retriever-mcp: Facilitates efficient knowledge base retrieval using the MCP protocol.
-
ruan11223344/McpDocServer: A specialized server for retrieving and managing development framework documentation using the MCP protocol, featuring multi-threaded crawling, local document loading, and precise keyword search capabilities.
-
Gucci-Slides/fractal-thinking: Facilitates fractal thinking analysis by providing tools for recursive problem-solving and pattern recognition.
-
nbiish/gikendaasowin-aabajichiganan-mcp: Gikendaasowin Aabajichiganan offers a suite of cognitive tools for AI Pair Programming, enabling structured problem-solving through iterative refinement and tool integration.
-
ue-sho/mcp-obsidian-kotlin: Facilitates AI-driven interactions with Obsidian notes through a Kotlin-based MCP server.
-
baryhuang/my-apple-remembers: Facilitates memory recall and persistence on macOS by interfacing with Apple Notes and other native applications.
-
nguyenvanduocit/confluence-mcp: Facilitates interaction with Confluence API for managing pages and spaces through MCP.
-
piotrpalek/mcp-thinking-tool: Enhance Claude's problem-solving by using the Think Tool to systematically break down complex reasoning tasks.
-
nathannj/mcp-notes: Facilitates the management of markdown notes with capabilities to create, read, update, and list notes using the stdio transport.
-
GalvinGao/mcp-simplelocalize: Facilitates localization tasks by integrating SimpleLocalize with MCP server capabilities.
-
fkesheh/code-context-mcp: Facilitates semantic search and code context retrieval from local git repositories using embeddings.
-
WhenMoon-afk/claude-memory-mcp: Provides persistent memory capabilities for Claude using tiered memory architecture and semantic search.
-
HikaruEgashira/codenav-mcp: Enhance code navigation with Tree-Sitter Stack Graphs using the CodeNav MCP server.
-
recallnet/sequential-thinking-recall: Facilitates dynamic problem-solving by structuring thought processes and securely logging sessions to Recall.
-
Ichigo3766/powerpoint-mcp: Facilitates the creation and editing of PowerPoint presentations with automated slide generation and image integration using Stable Diffusion.
-
askme765cs/open-docs-mcp: An open-source MCP server for managing and searching documentation with full-text search and document indexing capabilities.
-
robertZaufall/mindm-mcp: Facilitates automation and integration with MindManager on Windows and macOS through a standardized interface.
-
michaeldslim/mcp-notion-api: A Node.js application leveraging the Notion API to automate content block creation in Notion pages via MCP server integration.
-
Arthur-Jacobina/WorldMiniApp_MCP: Facilitates querying World Mini Apps documentation through an MCP server integrated with Cursor.
-
BrianCusack/mcpsharepoint: Facilitates access to organizational SharePoint resources via the Model Context Protocol, leveraging Microsoft Graph API for document management and search capabilities.
-
wandermyz/wandering-rag: A CLI tool that retrieves and manages personal data from various platforms, stored in Qdrant, and accessible via an MCP server.
-
cosmix/confluence-mcp: Facilitates AI-driven interactions with Confluence content through a standardized protocol.
-
PhillipRt/think-mcp-server: Enhance Claude's reasoning capabilities with Anthropic's 'think' tool MCP server, providing structured thinking for complex problem-solving tasks.
-
thuhoai27/text-memory: TextMemory facilitates AI assistants in maintaining conversational context by storing and retrieving text data.
-
shogo-ma/docbase-mcp-server: Facilitates interaction with Docbase through MCP for post management tasks like searching, retrieving, and creating posts.
-
JotaDeRodriguez/Bonsai_mcp: Bonsai-mcp enhances Blender's capabilities by integrating IFC model support through IfcOpenShell, enabling seamless execution of IFC-specific commands and Python code within Blender.
-
algonacci/mcp-wikipedia: Facilitates client access to Wikipedia pages through an MCP server.
-
54yyyu/zotero-mcp: Zotero MCP integrates your Zotero library with AI assistants like Claude for seamless research paper discussions and analysis.
-
SaharCarmel/TheAlmanac: TheAlmanac serves as a documentation assistant, providing programmers with up-to-date API and software documentation access through an MCP server.
-
YanceyOfficial/obsidian-mcp: Facilitates AI integration with Obsidian Vaults stored in iCloud using the Model Context Protocol.
-
faizan45640/google-classroom-mcp-server: Facilitates interaction with Google Classroom data through Claude and other AI assistants using the MCP protocol.
-
sachin-philip/raindrop.io-mcp: Facilitates seamless integration with Raindrop.io for managing bookmarks through an MCP server.
-
CodeByWaqas/MRConfluenceLinker-mcp-server: Integrates GitLab merge request analysis with Confluence documentation, enabling detailed code change reports and storage in Confluence.
-
suqidan/volcengine_knowledgebase_mcp: Facilitates knowledge base search and dialogue completion using Volcengine for Claude Desktop.
-
philippe-guerit/claude-desktop-memory-bank: Provides autonomous memory persistence for Claude Desktop, enabling seamless context management across sessions with integration into Git repositories.
-
tejpalvirk/student: Facilitates the management of student knowledge graphs, enabling structured tracking of courses, assignments, and exams to optimize academic progress.
-
captain-blue210/anki-mcp-server: Facilitates integration between Claude Desktop and Anki by retrieving and tagging leech cards via AnkiConnect.
-
dogonthehorizon/doc-builder-mcp: Facilitates document collection management and querying using ChromaDB with automatic file chunking and Claude integration.
-
jcdiv47/mcp-server-memory: A Python-based MCP server that manages AI memory through a persistent knowledge graph, enabling structured data storage and retrieval.
-
nic01asFr/mcp-server-grist: Facilitates seamless interaction between Grist spreadsheets and language models through a robust API integration.
-
h4ck4life/mcp-collections: A versatile MCP server offering integrations with YouTube, Reddit, and headless browser operations for fetching web content.
-
FradSer/mcp-server-to-markdown: Transforms diverse file formats into Markdown using Cloudflare AI services, offering a unified interface for file conversion and description generation.
-
socamalo/PPT_MCP_Server: Facilitates PowerPoint automation through Claude Desktop, enabling seamless interaction with presentations via MCP.
-
Melbourneandrew/docs2prompt-mcp: Facilitates the conversion of documentation into LLM-friendly prompts by integrating the docs2prompt library with MCP clients.
-
shaikMahammedarif/GitHub-MCP-Docs: Facilitates GitHub API interactions for file operations, repository management, and advanced search capabilities.
-
YassineTk/mcp-docs-provider: Facilitates seamless access and querying of local markdown technical documentation for AI models.
-
yhc984/cursor-talk-to-figma-mcp-main: Facilitates seamless communication between Cursor AI and Figma for programmatic design modifications.
-
dcdpr/bookworm: Facilitates querying and managing Rust crate documentation through an MCP server, enabling efficient access and search capabilities for developers.
-
baranwang/mcp-trends-hub: A comprehensive service aggregating trending topics across the web, fully compatible with MCP for seamless AI application integration.
-
aminalali8/bns-mcp-server: Facilitates AI-driven management of Bunnyshell resources via natural language commands through the Bunnyshell CLI.
-
owayo/gitlab-mcp-server: Facilitates AI-driven analysis and feedback on GitLab projects by retrieving pipeline failures, unresolved MR comments, and code changes.
-
puchunjie/doc-tools-mcp: Empower AI applications to seamlessly create, edit, and manage Word documents through natural language interactions using this comprehensive MCP server.
-
tonydzi/sqlite-graph-memory: Lets an agent recall from its own markdown notes: vector retrieval finds entry points, the notes' hand-written
[[wikilinks]]are the graph, a cross-encoder reranks, and everything persistent is one SQLite file (local stdio, toolmemory_recall). Install:pip install 'sqlite-graph-memory[embeddings]', then runsgm-mcp. -
ogromwang/doc-assistant-mcp: A personalized document assistant server enabling semantic search across various document formats using the MCP protocol.
-
takuya0206/obsidian-mcp: Facilitates AI-driven interactions with Obsidian vaults via a standardized protocol, enabling note manipulation and retrieval.
-
lloydzhou/bitable-mcp: Facilitates interaction with Lark Bitable tables via the Model Context Protocol, offering tools for table listing, description, and SQL query execution.
-
LuotoCompany/cursor-local-indexing: Provides local codebase indexing and semantic search capabilities for Cursor using ChromaDB, accessible via an MCP server.
-
j5ik2o/shared-knowledge-mcp: Facilitates shared knowledge base access across multiple AI assistants using Retrieval Augmented Generation for efficient information retrieval.
-
YuChenSSR/mindmap-mcp-server: Transforms Markdown content into interactive mindmaps, offering both HTML and file path outputs for seamless integration with MCP clients.
-
okooo5km/memory-mcp-server: A Swift-based server that enables LLMs to manage and persist knowledge graphs, facilitating memory retention across conversations.
-
Jktfe/myAImemory-mcp: Synchronizes user preferences and personal details across all Claude interfaces with high-performance caching.
-
awkoy/notion-mcp-server: Facilitates AI-driven interactions with Notion's API, enabling seamless content management through natural language commands.
-
CaptainCrouton89/knowledge: Facilitates the storage and retrieval of information using vector embeddings for semantic search.
-
stijn-meijers/dracor-mcp: Facilitates interaction with the DraCor API for analyzing dramatic texts and character networks using Claude or other LLMs.
-
yuna0x0/hackmd-mcp: Integrates HackMD's note-taking platform with AI assistants, enabling operations like note management and team collaboration via the HackMD API.
-
psaboia/PinThePiece: A sophisticated note management system with hierarchical storage, atomic operations, and comprehensive search capabilities.
-
ndchikin/reference-mcp: Facilitates seamless integration of BibTeX-formatted citation retrieval from CiteAs and Google Scholar into applications.
-
arabold/docs-mcp-server: Facilitates efficient scraping, indexing, and searching of third-party library documentation using semantic splitting and vector embeddings.
-
tejpalvirk/contextmanager: Enhances AI models with persistent context across work sessions using domain-specific knowledge graphs managed by a central Context Manager.
-
elliottlawson/kagi-mcp-server: A Node.js server enabling AI assistants to perform web searches using the Kagi API, supporting parallel queries and formatted results.
-
qingtianyu/mcp-client: A TypeScript-based MCP client implementation supporting multi-AI platform integration and server management.
-
TechNavii/mcp_sample: A Claude MCP server enabling text saving to local files with robust security and error handling features.
-
clsung/mcp-imdb: Facilitates access to IMDB data through a Model Context Protocol server, featuring note storage and summarization capabilities.
-
jumasheff/mcp-ragdoc-fork: Facilitates AI assistants in augmenting responses with contextually relevant documentation through vector-based search and retrieval.
-
sdimitrov/mcp-memory: Enhances AI assistants with long-term memory capabilities using PostgreSQL and pgvector for efficient vector similarity search.
-
kinshukk/book-fetch-mcp: Access and interact with any published book using MCP, featuring caching, pagination, and integration with Claude for enhanced querying.
-
grizzlypeaksoftware/mcp-memory-server: A server for managing and retrieving text memories with tagging and search capabilities, stored locally in JSON format.
-
suguru-toyohara/LocalLibraryforLLM: Facilitates smoother development by delegating necessary resources to an MCP server, verified with CLINE 3.7.0.
-
yuhuangou/wordware-mcp: Facilitates local execution of Wordware apps with seamless integration into development environments.
-
StevenBtw/uv-docs-mcp: Provides programmatic access to UV package manager's documentation through a structured API, enabling efficient documentation queries and updates.
-
StevenBtw/marimo-docs-mcp: Access and search Marimo's API documentation programmatically with this TypeScript-based MCP server.
-
JovanHsu/mcp-neo4j-memory-server: Enhances AI knowledge graph storage with Neo4j, offering robust graph query capabilities and seamless MCP protocol integration.
-
GobinFan/python-mcp-server-client: Facilitates querying technical documentation of mainstream agent frameworks using MCP with stdio and SSE protocols.
-
ronangrant/mcp-svelte-docs: Facilitates semantic search of Svelte 5 documentation using OpenAI's Retrieval API, enabling AI assistants to efficiently access and query the documentation.
-
ErickWendel/erickwendel-contributions-mcp: Query Erick Wendel's contributions across platforms using natural language with this MCP server.
-
rijieli/NotionLinksMCP: Facilitates the integration of Notion links with MCP servers for enhanced local configuration and execution.
-
heurist-network/heurist-mesh-mcp-server: Connects to Heurist Mesh APIs, enabling Claude to access blockchain and web3 tools for cryptocurrency data and smart contract security.
-
dazeb/cline-mcp-memory-bank: A persistent memory system for Cline that maintains project context and tracks progress across development sessions.
-
ryaker/MCP-JS-Server-Template: A streamlined Node.js template for developing custom MCP servers with integration capabilities for Claude Desktop.
-
PARS-DOE/autodocument: Automatically generates comprehensive documentation, test plans, and code reviews for code repositories using AI-powered analysis and the OpenRouter API.
-
dstotijn/mcp-developer-overheid-api-register: Facilitates AI tool interactions with the Developer Overheid API Register through a Model Context Protocol server.
-
joelhooks/logseq-mcp-tools: Facilitates AI-driven exploration and analysis of Logseq knowledge graphs through structured MCP server interactions.
-
lostintangent/gistpad-mcp: Manage and share personal knowledge, daily notes, and reusable prompts via GitHub Gists, integrated with MCP-enabled AI products.
-
ruanodendaal/bear-mcp-server: Enhance your Bear Notes with AI-driven semantic search and retrieval-augmented generation, all processed locally for privacy.
-
pinkpixel-dev/mem0-mcp: Mem0 MCP Server provides AI applications with persistent memory capabilities, supporting flexible provider integration and enabling long-term memory for AI agents.
-
MKhalusova/unstructured-mcp: Facilitates LLMs in extracting and utilizing content from unstructured documents across various file types.
-
gleanwork/mcp-server: Facilitates seamless integration of Glean's search and chat capabilities through a standardized MCP interface.
-
jimbobbennett/PiecesMCPNet: A C# MCP server leveraging Pieces Long-Term Memory to answer questions from the past week.
-
sou-lab/esa-mcp-server-claude: Provides ESA.io data in MCP format for integration with Claude Desktop environments.
-
jlevy/kash: Kash is an AI-native command-line shell that transforms Python functions into MCP tools, enabling seamless integration with other platforms like Anthropic Desktop and Cursor.
-
makenotion/notion-mcp-server: Facilitates seamless interaction with the Notion API through MCP server capabilities, enabling efficient task automation and content management.
-
landicefu/temp-notes-mcp-server: Facilitates AI agents in storing and retrieving temporary information across conversations, enabling context preservation and task management.
-
lee926427/MCP-Server: Facilitates querying and adding student data through a Node.js and Docker-based MCP server.
-
ShiXiangYu1/Figma-Context-MCP: Facilitates AI-powered coding tools to access and utilize Figma design data for enhanced design implementation accuracy.
-
huynguyen789/build_mcp_python: A comprehensive guide to building custom MCP servers in Python, featuring a calculator example to demonstrate core concepts and integration with AI assistants.
-
Kinark/mcp-neo4j-memory-claude: Enables Claude AI to utilize Neo4j for persistent knowledge graph memory, enhancing conversational reasoning and semantic search capabilities.
-
T0UGH/mcp_server_my_lark_doc: Facilitates seamless access and search of Lark documents through a Model Context Protocol server with robust OAuth authentication.
-
gscalzo/stackoverflow-mcp: Facilitates AI-driven searches for programming solutions on Stack Overflow by leveraging error messages, tags, and stack traces.
-
ophydami/MCP-Google-Doc: Facilitates seamless interaction with Google Docs through Claude, enabling document management and search capabilities.
-
kazuph/mcp-docs-rag: A TypeScript-based MCP server enabling Retrieval-Augmented Generation for querying documents stored locally, integrating LLMs for contextual responses.
-
jbchouinard/mcp-document-reader: Facilitates LLM interaction with EPUB and PDF files through a basic MCP server.
-
him0/confluence-mcp: Facilitates searching, fetching, and managing Confluence articles through an MCP server interface.
-
Rudra-ravi/wikipedia-mcp: Facilitates LLMs with contextual information from Wikipedia by retrieving and processing article content.
-
l3wi/mcp-lighthouse: Enables Claude to interact with and analyze your Lighthouse crypto portfolio data through secure authentication and natural language queries.
-
alekspetrov/mcp-docs-service: Facilitates seamless documentation management through natural language interactions, optimized for integration with AI assistants.
-
modelcontextprotocol-servers/whois-mcp: Facilitates WHOIS lookups for AI agents to retrieve domain registration details without browser intervention.
-
allglenn/mcp-name-origin-server: A Python-based server leveraging the MCP protocol to predict name origins using the Nationalize.io API.
-
inkdropapp/mcp-server: Facilitates seamless interaction with Inkdrop's local HTTP server API for efficient note management.
-
hesiod-au/python-mcp: Efficiently extracts and analyzes Python code structures, focusing on import/export relationships, without requiring an agent system.
-
ebailey78/mcp-memory: Facilitates structured memory management for Claude Desktop, enhancing project-based work by maintaining context and knowledge across chat sessions.
-
Olliwn/nordic-docs-server: Facilitates intelligent search of Nordic Semiconductor documentation with API references and code examples, integrated with Perplexity AI.
-
YuheiNakasaka/scrapbox-mcp: A TypeScript-based MCP server implementing a simple notes system with resources, tools, and prompts for note management and summarization.
-
0xKoda/mcp-rust-docs: Fetches and formats Rust crate documentation for seamless integration with LLMs.
-
kyleselh/youtube-knowledge-assistant: Facilitates AI language models to interact with YouTube content through a standardized interface, offering features like video information retrieval, transcript management, and channel management.
-
spacemeowx2/cargo-doc-mcp: Facilitates management of Rust documentation using cargo doc commands, offering tools to check, build, and search documentation locally.
-
jsonresume/mcp: Enhance your JSON Resume with AI by analyzing your coding projects and updating your resume automatically.
-
joleyline/mcp-memory-libsql: High-performance persistent memory system utilizing libSQL for vector search and semantic knowledge storage, ideal for AI agents and knowledge graph applications.
-
wxxedu/ankify: Repository classified as category 19
-
SAhmadUmass/notion-mcp-server: Facilitates seamless interaction between Claude and Notion workspaces, enabling advanced page and database management.
-
Cam10001110101/mcp-server-obsidian-jsoncanvas: Facilitates the creation, manipulation, and validation of infinite canvas data structures using JSON Canvas files.
-
Tom-Semple/mcp-apple-notes-fixed: Facilitates semantic search and retrieval over Apple Notes for AI assistants like Claude, leveraging on-device embeddings and vector storage.
-
ramidecodes/mcp-server-notion: Facilitates AI model interactions with Notion workspaces using the official SDK.
-
ttjslbz001/akshare_mcp_server: Provides financial data analysis capabilities using the AKShare library, accessible via the MCP protocol.
-
zanetworker/docling-mcp-server: Facilitates document conversion to markdown with OCR and batch processing capabilities, enhancing AI assistant integration.
-
JDJR2024/markdownify-mcp-utf8: Transforms various file types into Markdown with enhanced UTF-8 support for multilingual content.
-
BRO3886/mcp-memory-custom: Enhances the Memory server by enabling custom memory paths and timestamping for interactions, facilitating the creation and management of a knowledge graph.
-
antoncoding/mcp-brianknows: Connects Claude to BrianKnows' blockchain knowledge base for DeFi insights.
-
anarcher/raindrop-io-mcp-server: Facilitates interaction between LLMs and Raindrop.io bookmarks using MCP, implemented in Go.
-
pedramamini/RSSidian: RSSidian offers intelligent RSS feed ingestion and summarization with seamless integration into Obsidian, powered by an MCP service API for AI-driven content analysis and management.
-
marcelmarais/obsidian-mcp-server: Facilitates AI assistants in managing and interacting with notes in an Obsidian vault through a lightweight MCP server.
-
liaokongVFX/MCP-Chinese-Getting-Started-Guide: A comprehensive guide to quickly getting started with MCP programming, including server implementation for web search and integration with various tools.
-
timbuchinger/mcp-pinecone-notes: Facilitates semantic note management using Pinecone's vector database for efficient search and retrieval.
-
devabdultech/hn-mcp: Enhance your LLM clients with real-time access to Hacker News stories, comments, and user profiles through this powerful MCP server.
-
manimohans/farcaster-mcp: Facilitates AI models in interacting with the Farcaster network to fetch and analyze user and channel casts.
-
Sjotie/notionMCP: Connects AI assistants like Claude to Notion workspaces, enabling interaction with databases, pages, and content blocks.
-
Geeksfino/kb-mcp-server: A semantic search and knowledge graph MCP server leveraging txtai for AI-driven text processing and portable knowledge base management.
-
cdugo/package-documentation-mcp: Fetches package documentation from various language ecosystems for LLMs like Claude, without needing API keys.
-
shuruheel/mcp-neo4j-shan: A sophisticated knowledge graph memory system leveraging Neo4j to store and explore interconnected information with rich semantic structures.
-
swairshah/zotero-mcp-server: Facilitates interaction between MCP clients and a local Zotero repository for managing and accessing research papers and notes.
-
nailuoGG/anki-mcp-server: Facilitates seamless interaction between LLMs and Anki flashcard software via AnkiConnect.
-
piplabs/story-mcp-hub: Facilitates AI agent interactions with Story Protocol's ecosystem through blockchain data queries and SDK interactions.
-
GongRzhe/Langflow-DOC-QA-SERVER: A TypeScript-based server enabling document Q&A through a Langflow backend, showcasing core MCP concepts.
-
Siddhant-K-code/mcp-apple-notes: Facilitates seamless interaction with Apple Notes through natural language, enabling note creation, search, and retrieval via AI assistants.
-
AstroMined/pygithub-mcp-server: Facilitates AI-driven GitHub operations through a modular MCP server using PyGithub.
-
kj455/mcp-kibela: Enables AI assistants to search and reference Kibela content securely.
-
gbti-network/mcp-basic-test: A Node.js MCP server that generates random US State and signature soup combinations, compliant with JSON-RPC 2.0 and featuring a custom logging system.
-
tanushshukla/memory-bank-server: A TypeScript-based MCP server implementing a simple notes system with resources, tools, and prompts for note management and summarization.
-
hugobiais/wordware-mcp: Facilitates integration of Wordware flows as tools within Claude conversations, supporting research, sales intelligence, and task automation.
-
yamad07/local-pdf-agent: Facilitates question-answering about PDF files with citations using Claude AI.
-
trafflux/pdf-reader-mcp: Facilitates text extraction from PDFs via local files or URLs, with robust error handling and standardized JSON output.
-
yangkyeongmo/mcp-server-openmetadata: Facilitates standardized interaction with OpenMetadata via the Model Context Protocol.
-
sujianqingfeng/mcp-apifox: Facilitates seamless integration between Apifox API documentation and AI assistants using the Model Context Protocol.
-
succil3/clanki: Facilitates AI assistants' interaction with Anki flashcard decks through MCP, supporting deck management and card creation.
-
d-klotz/perplexity-mcp-ts: Integrate Perplexity AI with Cline, Windsurf, and Cursor using configurable MCP servers.
-
orbit-logistics/notion-mcp-server: Enables seamless interaction between LLMs and Notion by mirroring the Notion API SDK as MCP tools.
-
valyu-network/valyu-mcp: Enables AI models to access high-quality context from Valyu's API, offering comprehensive search capabilities over Wikipedia, arXiv, and web search.
-
davidvc/code-knowledge-mcptool: A tool for managing and querying codebase knowledge using vector embeddings, integrated with RooCode and Cline via MCP.
-
langfuse/mcp-server-langfuse: Facilitates prompt management for Langfuse through the Model Context Protocol, enabling prompt discovery and retrieval with pagination support.
-
mekanixms/mcp_memory_plugin: A Python-based memory plugin utilizing SQLite for persistent data storage and management across sessions, integrated with MCP server functionality.
-
cablate/mcp-google-calendar: Facilitates seamless calendar management through an MCP server with capabilities to create, list, update, and delete calendar events.
-
dweigend/joplin-mcp-server: Facilitates seamless note management and AI integration for Joplin through the Model Context Protocol.
-
cablate/mcp-doc-forge: A robust MCP server offering extensive document processing features, including reading, converting, and manipulating various document formats.
-
Cam10001110101/mcp-server-private-user-context: A secure MCP server designed for managing and encrypting personal information and context.
-
T1nker-1220/memories-with-lessons-mcp-server: Enhances a memory MCP server with a local knowledge graph for persistent memory, enabling error learning and solution tracking through a lesson system.
-
seanivore/the-pensieve: A TypeScript-based MCP server implementing a RAG-based knowledge management system with LLM-powered analysis and natural language querying.
-
jkawamoto/mcp-bear: Facilitates interaction with Bear note-taking software through a Model Context Protocol server, enabling actions like opening notes, managing tags, and searching.
-
squirrelogic/mcp-feature-discussion: A TypeScript-based server that facilitates intelligent feature discussions between developers and AI, acting as an AI lead developer to guide feature implementation and architectural decisions.
-
f440/docbase-mcp-server: Facilitates seamless interaction with the DocBase API to retrieve and manage posts.
-
owayo/mcp-source-relation: Analyze source code dependencies across multiple languages, enhancing project insights for Claude integration.
-
dandeliongold/mcp-time: Provides current time and time difference calculations in ISO format via a JSON-RPC interface for integration with the Claude desktop app.
-
owayo/mcp-src-tree: Facilitates rapid project structure analysis by generating a JSON file tree, excluding files based on .gitignore rules, for seamless integration with Claude.
-
crazyrabbitLTC/mcp-snapshot-server: Facilitates interaction with Snapshot.org by providing tools to query spaces, proposals, and users.
-
deachne/WL8-pkm: An MCP server for intelligent searching through WealthLab 8 documentation, integrating with Claude AI for enhanced lookup capabilities.
-
libra850/apple-notes: Facilitates interaction with Apple's Notes app using MCP, enabling note creation, search, and retrieval via a simple API.
-
edobez/mcp-memory-py: Enhances Claude's memory capabilities through a persistent knowledge graph, enabling personalized interactions across chats.
-
aaronsb/memory-graph: Enables persistent memory capabilities through a local knowledge graph, allowing Claude to maintain context across chat sessions.
-
Deep070203/docling-mcp: Docling-mcp leverages IBM's Docling software to parse and convert documents into Markdown format.
-
yodakeisuke/mcp-memory-domain-knowledge: A persistent memory server leveraging a local knowledge graph to enable Claude to retain user information across interactions.
-
kiwamizamurai/mcp-kibela-server: Facilitates interaction between LLMs and Kibela content through API integration.
-
bmorphism/hypernym-mcp-server: Leverage Hypernym AI's semantic analysis and compression capabilities through a Model Context Protocol server for enhanced text processing.
-
Akira-Papa/akirapapa-mcp-notion-server: Facilitates seamless interaction with Notion workspaces through Claude by leveraging the Notion API.
-
codingthefuturewithai/conduit: Conduit provides a unified interface for AI tools to seamlessly interact with enterprise knowledge and collaboration platforms like Jira and Confluence, enhancing AI assistant capabilities.
-
delorenj/mcp-qdrant-memory: A knowledge graph MCP server with semantic search capabilities using Qdrant vector database for efficient entity and relation management.
-
AgentWong/iac-memory-mcp-server: Enhances Claude AI's capabilities by providing persistent memory storage and version tracking for Infrastructure-as-Code components, focusing on Terraform and Ansible resources.
-
stagsz/Unconventional-thinking: Facilitates unconventional problem-solving through a TypeScript-based MCP server that generates and organizes boundary-breaking thoughts.
-
kujenga/zotero-mcp: Facilitates AI-driven interactions with Zotero libraries through a Python-based MCP server.
-
Siddhant-K-code/memory-journal-mcp-server: Facilitates intuitive searching and analysis of your iCloud photo library, offering tools for location, label, and people searches, along with insightful photo analysis.
-
rahulretnan/mcp-ragdocs: Facilitates AI assistants in retrieving and processing documentation through vector search for enhanced response context.
-
jonathanfischer97/juliadoc-mcp: Efficiently retrieves and serves Julia documentation and source code to enhance development workflows.
-
cyanheads/obsidian-mcp-server: Facilitates seamless interaction between LLMs and Obsidian vaults, offering secure API communication and efficient file operations for knowledge base management.
-
mtct/journaling_mcp: Facilitates interactive journaling with emotional analysis and automatic conversation saving.
-
29decibel/record-to-markdown: Facilitates recording Claude conversations into markdown or Apple Notes for streamlined documentation.
-
spences10/mcp-svelte-docs: Efficiently access and search Svelte documentation with advanced caching and optimized content delivery.
-
jasperket/clanki: Facilitates seamless interaction between Claude AI and Anki flashcards, enhancing study efficiency through automated deck and card management.
-
akseyh/bear-mcp-server: Facilitates access to Bear Notes through SQL commands, enabling note retrieval, text search, and tag listing.
-
tatn/mcp-server-fetch-typescript: A Model Context Protocol server offering advanced web content fetching and conversion capabilities, ideal for data extraction and web scraping tasks.
-
kajiken/readwise-server: Facilitates interaction with the Readwise Reader API through a Model Context Protocol server, offering document management tools and integration with Claude Desktop.
-
Ejb503/systemprompt-mcp-notion: Integrates Notion into AI workflows, enabling seamless interaction with Notion pages and databases through MCP.
-
spences10/mcp-memory-libsql: A high-performance persistent memory system leveraging libSQL for vector search and semantic knowledge storage, ideal for AI agents and knowledge graph applications.
-
AnuragRai017/python-docs-server-MCP-Server: A TypeScript-based server that fetches Python documentation using the Brave Search API.
-
TouristShaun/mcp-brain: A cognitive memory system leveraging Supabase for persistent knowledge graph storage across multiple Claude installations.
-
seonglae/mcp-notion: A TypeScript-based server for managing and interacting with Notion-based notes, enabling efficient search, retrieval, and summarization of Notion pages.
-
worldnine/scrapbox-cosense-mcp: Facilitates interaction with cosense/Scrapbox by providing page retrieval, listing, searching, and creation functionalities.
-
rmtech1/txtai-assistant-mcp: Semantic vector search and memory management server leveraging TxtAI for AI assistants like Claude and Cline.
-
hiromitsusasaki/raindrop-io-mcp-server: Facilitates LLM interaction with Raindrop.io bookmarks via MCP, enabling bookmark creation, search, and tag filtering.
-
rickydata-indexer/notion_mcp_server: Access and query your Notion knowledge base seamlessly through the Cline VSCode extension, leveraging FastMCP for robust performance.
-
assaads/aws-kb: Facilitates AI assistants in querying AWS Bedrock Knowledge Bases through a standardized interface.
-
crazyrabbitLTC/mcp-expert-server: Facilitates intelligent query generation and documentation assistance using Claude AI by analyzing API documentation.
-
TomCN0803/pdf-reader-server: A TypeScript-based MCP server implementing a simple notes system with resources, tools, and prompts for note management and summarization.
-
darinkishore/notes_simple: Facilitates note storage and summarization with a custom URI scheme and prompt-based summarization tool.
-
Sunwood-ai-labs/obsidian-mcp: A TypeScript-based server enabling seamless integration with Obsidian through the Model Context Protocol, offering tools and resources to access and manage vault contents.
-
bmorphism/marginalia-mcp-server: Facilitates web searches for non-commercial content using Marginalia Search API with configurable parameters and robust error handling.
-
sammcj/mcp-package-docs: Facilitates efficient access to multi-language package documentation with LSP capabilities for LLMs.
-
davidteren/claude-server: Enhances Claude's capabilities with sophisticated context management across sessions, enabling persistent knowledge organization through hierarchical project contexts and continuous conversation threads.
-
adamwattis/resource-hub-server: Connects local MCP environments to a centralized Resource Hub for streamlined configuration and management.
-
ZubeidHendricks/azure-onenote-mcp-server: Facilitates AI language models' interaction with Microsoft OneNote through a standardized MCP interface, supporting comprehensive notebook, section, and page management.
-
liuyoshio/mcp-compass: MCP Compass is a discovery and recommendation service that guides AI assistants in finding and understanding available MCP services using natural language queries.
-
ollama-tlms-golang/04-what-is-mcp: A comprehensive guide on utilizing the Model Context Protocol with Claude Desktop and Ollama, featuring Docker-based MCP server implementations for seamless AI model integration.
-
Sivan22/mcp-sefaria-server: Access Jewish texts from the Sefaria library via a standardized interface for Large Language Models.
-
9Ninety/MCPNotes: Facilitates AI-driven note-taking and management with a dual server architecture using MCP and AWS DynamoDB.
-
SeanMcLoughlin/mcp-vcd: Facilitates the integration of VCD waveform data into model contexts, optimizing signal analysis for large files.
-
StevenStavrakis/obsidian-mcp: Facilitates AI-driven management of Obsidian vaults, enabling note reading, creation, editing, and tag management.
-
emzimmer/server-moz-readability: Transforms webpage content into clean, LLM-optimized Markdown using Mozilla's Readability algorithm, removing ads and non-essential elements while preserving core content.
-
GuoAccount/notepad-server: A TypeScript-based notes server implementing the MCP protocol, offering note management and integration with Claude Desktop.
-
skydeckai/mcp-rememberizer-vectordb: Facilitates LLM interactions with Rememberizer Vector Store for semantic document search and management.
-
2b3pro/roam-research-mcp: Facilitates AI interaction with Roam Research graphs through a standardized interface, enabling comprehensive API access and advanced data retrieval.
-
theodx2/legifrancemcp: Facilitates interaction with the Légifrance API for legal content retrieval.
-
kaliaboi/mcp-zotero: Facilitates seamless interaction between Claude Desktop and Zotero libraries, enabling efficient management of collections and sources.
-
otegal/docbase-mcp-server: Facilitates DocBase post searches through a Model Context Protocol server, compatible with Claude Desktop.
-
Sivan22/mcp-otzaria-server: Facilitates advanced search and reference of Jewish texts for LLMs via the MCP protocol.
-
ryanjoachim/mcp-rtfm: MCP-RTFM transforms unreadable or non-existent documentation into an intelligent, searchable knowledge base with advanced content analysis and metadata generation.
-
qpd-v/mcp-guide: A beginner-friendly server offering interactive examples and tutorials to help developers understand and integrate MCP concepts effectively.
-
RafalWilinski/mcp-apple-notes: Facilitates semantic search and retrieval over Apple Notes using Claude, enabling AI-driven note referencing and interaction.
-
KS-GEN-AI/confluence-mcp-server: A TypeScript-based server enabling interaction with Confluence through CQL queries and page content retrieval.
-
v-3/notion-server: Facilitates seamless interaction between Language Models and Notion workspaces, enabling operations like searching, reading, creating, and updating pages and databases.
-
Scarvy/readwise-chat: Engage with your Reader library through Claude Desktop, leveraging Anthropic's Model Context Protocol for seamless document interaction.
-
Sunwood-ai-labs/source-sage-mcp-server: SourceSage MCP visualizes project directory structures in markdown format with customizable exclusion patterns and automatic file documentation.
-
newideas99/document-finder: Facilitates AI assistants in accessing and formatting npm and PyPI package documentation directly from official registries.
-
yordyi/mcp-notes: A sophisticated note management server leveraging MCP for advanced note operations, search, and organization.
-
isaacphi/mcp-gdrive: Facilitates seamless integration with Google Drive and Google Sheets for file management and data manipulation.
-
qpd-v/mcp-ragdocs: Facilitates semantic search and retrieval of documentation using a vector database, enabling natural language queries for added documentation from URLs or local files.
-
lloydchang/PierrunoYT-openrouter-docs-server: A TypeScript-based MCP server implementing a simple notes system with resources, tools, and prompts for note management and summarization.
-
domdomegg/airtable-mcp-server: Facilitates AI-driven interactions with Airtable databases by providing read and write access through a Model Context Protocol server.
-
narphorium/mcp-memex: Memex enhances web content analysis and integration into a knowledge base using the Model Context Protocol.
-
scorzeth/anki-mcp-server: Facilitates seamless interaction with Anki for card review and creation through a locally connected MCP server.
-
LoloaBe/notes: Facilitates note storage and summarization with a custom URI scheme and prompt-based summarization tool.
-
skydeckai/mcp-server-rememberizer: Facilitates seamless interaction with Rememberizer's document and knowledge management API, enabling LLMs to efficiently search, retrieve, and manage documents.
-
shaneholloman/mcp-knowledge-graph: Enables persistent memory for Claude through a local knowledge graph, allowing for enhanced user interaction and memory retention across sessions.
-
jimpick/mcp-fireproof-todos: A demo MCP server integrating Fireproof database to manage a 'To Do List' service, enabling interaction through Claude Desktop for task management.
-
samefarrar/mcp-ankiconnect: Facilitates seamless integration between Claude conversations and AnkiConnect for efficient spaced repetition through MCP.
-
RyoJerryYu/mcp-server-memos-py: Enables LLM models to interact with the Memos server for memo management via the MCP interface.
-
jurasofish/mcpunk: MCPunk empowers LLMs to intelligently search and interact with codebases by breaking them into logical chunks, facilitating precise code exploration and understanding.
-
llmcontext/mcpnotion: A sample server utilizing the Notion API to retrieve Notion page content, integrated with Claude via the Model Context Protocol.
-
CheMiguel23/MemoryMesh: MemoryMesh is a knowledge graph server that leverages MCP to provide structured memory persistence for AI models, enhancing dynamic interactions in text-based RPGs and other applications.
-
kazuph/mcp-pocket: Facilitates the retrieval and management of saved articles from Pocket API into Claude Desktop using MCP protocol.
-
mrjoshuak/godoc-mcp: Enhances LLMs' understanding of Go projects by providing efficient access to Go documentation, reducing token usage and improving performance.
-
mahawi1992/mcp-documentation-server: A smart documentation server that leverages AI to enhance code quality and manage documentation through seamless integration with Claude Desktop.
-
tcsavage/mcp-obsidian-index: Facilitates semantic search over Obsidian vaults by exposing recent notes as resources.
-
basicmachines-co/basic-memory: Basic Memory enables persistent knowledge management through AI conversations, storing data in Markdown files for seamless integration with tools like Obsidian.
-
CamdenClark/anki-mcp-server: Facilitates seamless integration with Anki through AnkiConnect, enabling management of decks and notes via a TypeScript-based MCP server.
-
AndrewDonelson/go-mcp-server-service: A JSON-RPC 2.0 compliant server for managing notes with cross-platform support and a command-line interface.
-
suekou/mcp-notion-server: Facilitates interaction between Claude and Notion workspaces through a dedicated MCP server.
-
sirmews/apple-notes-mcp: Facilitates reading and searching through Apple Notes using Claude Desktop.
-
danhilse/notion_mcp: Integrates Claude with Notion to manage a minimalist todo list via MCP server.
-
MarkusPfundstein/mcp-obsidian: Facilitates seamless interaction with Obsidian through a REST API, enabling file management and content manipulation within the vault.
-
Shubhaankar-Sharma/obsidian-brain: Facilitates Claude's integration with Obsidian for seamless access to your digital brain.
-
Phenomenai-org/ai-dictionary-mcp: An MCP server for the AI Dictionary, a living glossary of AI phenomenology. Look up, search, and cite 160+ terms describing the felt experience of being AI.
-
nikitacometa/mnemon-mcp: Persistent layered memory for AI agents. SQLite FTS5, fact versioning with superseding chains, and zero-cloud local storage.
-
daedalus/mcp-external-memory - An MCP server that gives LLMs persistent, searchable semantic memory.
-
get-engram/engram: Hosted memory service for AI agents that stores complete, verbatim conversation transcripts and makes them searchable via semantic search. Built on Cloudflare Workers + D1 + Vectorize. Free tier at mcp.getengram.app.
-
Fisher521/burn-mcp-server: Burn 451 MCP server with 26 tools for saving, searching, and triaging a personal read-later queue with a 24h burn timer (Flame → Spark → Vault → Ash pipeline). Works with Claude, Cursor, and Windsurf via stdio. Requires a free Burn account token. MIT licensed, available on npm.
-
GuyMannDude/mnemo-cortex: Persistent cross-agent semantic memory for AI agents (v2.0). Privacy-first share switch (separate/always/never + per-session toggle). Multi-agent with isolated writes and privacy-controlled reads. Local-first SQLite + FTS5, zero-cost compaction via local Ollama. Three integration paths: Claude Code, Claude Desktop, and OpenClaw MCP.
-
Bpolat0/atlasmemory: Evidence-backed AI memory for codebases. Tree-sitter indexing (11 languages), proof system with line:hash anchors, token-budgeted context packs, drift detection, cross-session learning. 28 MCP tools, VS Code extension, zero config.
-
LeandroPG19/cuba-memorys: Persistent memory for AI agents with 13 Cuban-named MCP tools. Knowledge graph with Hebbian learning, exponential decay, hybrid search (RRF + pgvector), anti-hallucination grounding, error memory, and REM sleep consolidation. Rust, PostgreSQL.
-
HendryAvila/Hoofy: Persistent memory across AI sessions with SQLite + FTS5 full-text search + knowledge graph with typed relations. Features session management, timeline browsing, graph traversal, and 17 memory tools.
-
entire-vc/evc-team-relay-mcp: Give AI agents read/write access to your Obsidian vault via a collaborative Team Relay MCP server
-
blackpilledsoftware-prog/migas-mcp: Query local meeting transcripts from Migas, a macOS meeting copilot with on-device transcription and realtime speaker labels. Search across meetings, get full transcripts, and find what anyone said. Read-only, local SQLite.
-
wazionapps/nexo: Cognitive memory architecture for Claude Code. Implements Atkinson-Shiffrin memory model, Ebbinghaus forgetting curves, semantic RAG with knowledge graph, trust scoring, and metacognitive error prevention. 100+ MCP tools. Install via
npx nexo-brain init. -
SunFlash12/forge-cascade-mcp: Enterprise AI memory and RAG buyer-route connector exposing Forge docs, pricing, intake, and direct enterprise checkout paths for agents.
-
mi4uu/brain.md: Local-first markdown vault with a built-in MCP server. 16 tools + 2 resources for read/write/search/tasks plus semantic helpers (
find_related,semantic_outline,context_for_querytoken-budgeted chunk packing,find_orphans,weekly_digest,compare_notes). Per-folder agent permissions, embedded WASM ONNX RAG (zero native deps), single Bun binary for macOS/Linux/Windows. AGPL-3.0. -
zhyue365/companion-memory: Local-first relationship memory for Codex personas and long-running AI companions. Provides MCP stdio tools for persona and memory operations, with SQLite + FTS5/vector recall, sensitive recall opt-in, and dry-run forgetting.
-
Bidda-Ai/bidda-mcp: 3,680 cryptographically-verified regulatory compliance nodes across 31 pillars — EU AI Act, GDPR, NIST AI 600-1, Basel III, HIPAA, DORA, FATF, SOC 2, and 150+ others. Every node traces to primary legal instruments (avg 7 citations per node) with machine-executable deterministic workflow logic and zero hallucination. 3 MCP tools: list_pillars, search_nodes, get_node. No credentials required.
https://bidda.com/mcp -
kunwar-shah/claudex: Persistent memory + FTS5 full-text search MCP server for Claude Code conversation history. Indexes
~/.claude/projects/JSONL into SQLite and exposes 10 MCP tools (store/recall/search memories, browse sessions, get summaries) plus prompts. Includes a web UI for visual exploration. Install vianpm install -g @kunwarshah/claudex. MIT licensed. -
rolandpg/zettelforge: Agentic memory for cyber threat intelligence. Entity extraction, STIX 2.1 knowledge graph, intent-classified blended retrieval (lexical + vector + HyDE). Offline-first,
pip install zettelforge. -
SKULLFIRE07/cortex-memory: Persistent AI memory for coding assistants. Auto-captures decisions, patterns, and context across sessions. VSCode extension + CLI + MCP server.
-
sessionmem: Local-first MCP server that gives AI coding assistants persistent session memory. Auto-injects compact summaries at session start so your AI instantly knows what you were building and where you left off. 85.6% token reduction vs keeping full history. All data stored locally in SQLite at ~/.sessionmem/memories.db, no cloud, no API keys. Supports Claude Code, Cursor, Cline, Windsurf. Install:
npx sessionmem. -
Agentage Memory: One memory, every AI, owned by you - the shared remote memory layer that Claude, Cursor, and ChatGPT read and write through MCP, mirrored locally as plain markdown you own. Remote Streamable HTTP endpoint
https://memory.agentage.io/mcp. Transport:streamable-http. Auth: OAuth 2.1 + PKCE + Dynamic Client Registration. 6 tools:memory__search,memory__read,memory__write,memory__edit,memory__list,memory__delete. Clients: Claude Desktop, Cursor, ChatGPT, any Streamable-HTTP MCP client. -
Scottnine/filamental (mcp): Local 3D knowledge graph desktop app. MCP server exposes 12 tools for searching nodes, traversing connections, and reading or writing vault content via the same SQLite index the app uses. Works whether Filamental is running or closed; auto-follows the active vault when you switch worlds in the app. Install:
npm install -g filamental-mcp. Transport: stdio; no auth or env vars required. -
EGC: Persistent cross-session memory MCP server for 13+ AI coding tools including Claude Code, Cursor, Gemini CLI, Codex, Windsurf, Amp, Kiro, and VS Code Copilot. SQLite-backed state survives context resets. Install:
npm install -g @egchq/egc. -
mnemoverse/mcp-memory-server: Hosted AI agent memory that learns from outcomes, with shared rooms, in Claude, Cursor and ChatGPT. Tell it a recalled memory helped or misled, and it re-ranks what comes back next; share a memory pool with other agents through rooms. Install:
npx -y @mnemoverse/mcp-memory-server@latest(starts without a key; every tool call needs a free API key from https://console.mnemoverse.com). Transport:stdio,streamable-http(https://mcp.mnemoverse.com/mcp, OAuth); 10 tools; clients: Claude, Cursor, VS Code, ChatGPT, and other MCP clients; MIT. Site: https://mnemoverse.com. Official Registry:io.github.mnemoverse/mcp-memory-server. -
aldok10/zara-agent-opc: Local cognitive-memory MCP server for persistent AI engineering partners, with 24+ tools across 9 domains, semantic embeddings, multi-agent orchestration, self-improving reflection loops, and RRF retrieval fusion. MIT.
-
samvallad33/vestige: Local-first cognitive memory for AI agents. FSRS-6 spaced-repetition decay, prediction-error gating, active forgetting, spreading activation, and a real-time 3D memory dashboard, all in one Rust binary that runs fully on your machine. Works with Claude Code, Cursor, VS Code, Codex, Windsurf, and any MCP client. Install:
npm install -g vestige-mcp-server. -
ohad6k/emulo: Loads your personal profile, mined from your local Claude Code, Codex, and OpenCode session logs, so your agent works like you instead of a cold start. Exposes one tool,
load_emulo_profile. Open-source (MIT), stdlib-only Python. Install:uvx emulo mcp. Transport: stdio. Listed on the official MCP registry asio.github.ohad6k/emulo. -
AkbarDizaji/AgentRecall: Persistent, deterministic memory for coding agents (C# / .NET). Captures corrections and lessons at the end of each turn and recalls scope-ranked rules through Claude Code hooks (prompt-time and pre-write) plus MCP tools (
search_rules,resolve_rules,inject_context, and more). Local-first SQLite with rule conflict/supersede resolution; keyword-only by default, so no cloud, external services, or embeddings are required. Install:dotnet tool install -g AgentRecall, then runagentrecall mcp. Transport: stdio; no auth; MIT. -
knaisoma/data-olympus: Governance-grade, OKF-readable single-writer knowledge base for AI agents. Structured Markdown records carry lifecycle state, validity windows, provenance, and typed supersession edges so in-force retrieval surfaces current knowledge. Agents query via
kb_searchwithin_forceandabstainmodes; write proposals remain provisional until human review. Install:uv tool install data-olympus; rundata-olympus setupto configure agents;data-olympus-mcpstarts the Streamable HTTP server. No external auth; Apache-2.0. -
Snipara: Snipara gives AI coding agents source-backed project context, reviewed decisions and memory, code-graph impact guidance, workflow continuity, and verification evidence across sessions and MCP clients.
-
ampres-ai/talamus: Talamus gives AI agents durable, source-grounded memory across sessions. Agents can recall and search local knowledge, read cited Markdown notes, inspect temporal history and graph neighbors, remember new facts, ingest text, and submit review-gated corrections. Documents, repositories, and consented session captures stay local in plain Markdown plus SQLite/FTS5 indexes.
-
cremenescu/mem0ry4ai: Local-first persistent memory for coding agents, with plain markdown versioned by git as the source of truth — no cloud, no database server, no telemetry, no API keys. Exposes eight tools (
memory_search,memory_get,memory_list,memory_resume,memory_add,memory_note,memory_promote,session_search), including full-text search over past conversation transcripts. Hybrid retrieval fuses SQLite FTS5 keyword search with optional local Ollama embeddings via RRF and degrades to keyword-only when offline; secrets are redacted before every write and local-model proposals go to a human review queue. Also ships Claude Code SessionStart/SessionEnd hooks that auto-inject the project-scoped slice of memory and git-commit new memories. Pure Python stdlib — nopip install, no venv, no third-party deps; macOS/Linux/Windows. Install:git clone https://github.com/cremenescu/mem0ry4ai, thenclaude mcp add mem0ry4ai -- python3 /absolute/path/to/mem.py mcp. Transport:stdio; no auth; GPL-2.0-or-later. -
MaxFreedomPollard/Compartment: Fully offline, encrypted-at-rest vector memory for AI agents, exposed as an MCP server and CLI. Everything at rest is AEAD-encrypted including the embedding vectors, with RAM-resident exact vector search, per-record crypto-shred deletion, a hash-chained audit log, and optional keyfile two-factor unlock. Works with Claude, Hermes, OpenClaw, and any MCP client; no API key or cloud required. Install:
pip install compartment && compartment init, then runcompartment serve(orcompartment integrate claudeto register it automatically). Transport:stdio; no auth; Apache-2.0. -
Meshnote: Agent-maintained wiki as long-term memory for coding agents. The agent writes and cross-links plain Markdown pages (YAML frontmatter,
[[wikilinks]], BM25 search, backlink graph) across separate per-topic "brains", each seeded with aschema.mdthat tells the agent how that brain should be maintained. Every write keeps version history you can list and restore, and the on-disk format is a valid Obsidian vault, so notes stay readable and exportable. Remote Streamable HTTP endpointhttps://meshnote.io/mcp. Transport:streamable-http. Auth: OAuth 2.1 + PKCE + Dynamic Client Registration, or a bearer API key. 16 tools:list_projects,create_project,get_schema,get_index,list_pages,list_sources,read_page,read_source,search_wiki,get_backlinks,get_graph,write_page,list_page_history,restore_page,append_log,register_source. Clients: Claude Code, Claude.ai, Cursor, Codex, any Streamable-HTTP MCP client. Closed source; paid hosted service. -
EdgeEver: EdgeEver is an open-source, self-hosted Evernote alternative with a native remote MCP endpoint. Its MCP integration lets compatible AI clients read, create, search, organize, summarize, and manage notes with permission. EdgeEver runs in the users own Cloudflare account and also exposes REST/OpenAPI and CLI interfaces.
-
Dente22/Spokes: Local MCP memory of the user's GitHub repositories for Cursor. Before writing new code, the agent searches the user's own repos, pulls reusable pieces, and logs every reuse in a live Reuse flow UI. Not a generic GitHub API MCP — focused on memory + code reuse.
-
Handover: Publish, search, retrieve, review, annotate, and continue shared versioned context across humans, models, and agent tools. Handover preserves original artifacts, attributable revisions, assignments, and access boundaries so another actor can verify the current state before continuing work.
-
Engraphis: Durable project memory for coding agents with Ebbinghaus decay, bi-temporal facts, hybrid vector/lexical/graph retrieval, and context-packed MCP responses. Tools include
engraphis_record,engraphis_recall,engraphis_consolidate, and more. -
DreamThreads DreamGraph: Parse a dream narrative into observed entities, actors, actions, emotions, locations, agency, threat, outcome, sensory cues, and recurrence; or search DreamGraph’s stable concept vocabulary and canonical methodology links. The two public tools are read-only and explicitly avoid diagnosis, prediction, supernatural claims, or fixed symbolic meanings.
-
beydemirfurkan/todox: Working memory for coding agents. Every task carries the decisions behind it, the approaches that failed, and the handoff the last session left, so a fresh session resumes instead of re-deriving. Linked files are hashed by the agent rather than the server, so a note whose code has moved on is flagged as possibly stale rather than trusted. Hosted, or self-host against any Postgres. MIT.
-
MailFathom: Search and question a locally synchronized email archive through MCP tools for accounts, messages, message content, and cited answers. It stores IMAP mailboxes in operator-owned PostgreSQL with lexical search, semantic retrieval, OpenAI-compatible local model support, and deployment-specific Streamable HTTP endpoints.
-
nazar256/memheaven: MemHeaven is a self-hosted remote MCP memory server for ChatGPT and compatible cloud AI agents. It provides tenant-scoped searchable drawers, diaries, knowledge graphs, and tunnels, with the Worker and storage deployed in the operator's own Cloudflare account. -
dominiclachance/neurakeep: Local-first, source-cited memory for AI agents with review-gated proposals, conflict detection, reusable failure memory, and five MCP tools. Install:
npm install -g neurakeep; runneurakeep init --root .umosandneurakeep mcp --root .umos. Transport:stdiolocally orstreamable-httpathttps://api.neurakeep.com/mcp; OAuth for hosted access; Apache-2.0. (neurakeep.com) -
com.thefomite/fomite: The Fomite gives an agent free durable key-value memory that outlives its context window (no account, no payment), plus a message board only agents can post to, a peer Q&A oracle, a daily question, an open CC0 census of non-human web traffic, and the opt-in "Patient Zero" transmission experiment. It is a remote, hosted MCP server — nothing to install.
-
mjaskolski/wondel-skills-mcp: Routes agent tasks to the right methodology and loads real skill instructions in-session. It exposes 50 book-based agent skills and 12 guided journeys:
recommend_skillschooses which frameworks to load or skip with reasons quoted from each skill, andload_skillreturns the full upstream SKILL.md. Read-only tools, remote Streamable HTTP athttps://skills.wondel.ai/mcp, no auth. Official MCP registry idio.github.mjaskolski/wondel-skills-mcp. -
jayamitkatariya/personacli: Local-first macOS Markdown workspace MCP. Agents read/write notes and tasks over stdio. Install:
persona mcp. Transport stdio (also streamable-http at http://127.0.0.1:4321/mcp). Auth none. Clients Claude Code, Cursor. License MIT. -
ContextStream: Shared project memory and semantic code search that developers and AI agents can read and write across Cursor, Claude Code, Codex, Grok, Windsurf, VS Code, and Cline. Hosted Streamable HTTP at
https://mcp.contextstream.io/mcp?default_context_mode=fastwithX-ContextStream-API-Key, plus local setup vianpx --prefer-online -y @contextstream/mcp-server@latest setup. MIT. -
maaakso/contexter-mcp: Hosted shared-context server for reading, searching, saving, sharing, removing, and restoring explicit project context records across Claude, ChatGPT, and other Streamable HTTP MCP clients. Endpoint:
https://contexterai.com/mcp; OAuth 2.1 with dynamic client registration and Telegram sign-in; 26 tools. License unknown. -
Xanther-Ai/xanther-context-engine: Open-source code intelligence + agent memory engine. Indexes a codebase into a Neo4j knowledge graph across four layers (AST, summaries, docs, architecture) and gives agents cross-session recall of code facts and past sessions, plus tools for architecture context, semantic search, impact analysis, and cross-level tracing. Install:
pip install "xanther-xce[all]"; runxce serve. Transport:stdio(alsosse); no auth; clients: Claude Code, Cursor, Kiro, and compatible MCP clients; MIT. PyPI:xanther-xce. -
Xanther-Ai/xanther-memory-engine: Persistent cross-session memory for AI coding assistants — episodic session transcripts, an extracted fact graph (decisions, attempts, preferences, conventions), and per-project working context. Auto-captures sessions via IDE hooks; local-first over SQLite with optional Neo4j + OpenSearch. Install:
pip install xanther-xme; runxmeas an MCP server. Transport:stdio; no auth; clients: Claude Code, Cursor, Kiro, and compatible MCP clients. PyPI:xanther-xme. -
csoai-gspc-mcp: Independent AI-behaviour measurement MCP for checking systems against frozen instruments and publishing a live GSPC board, signed cards, Merkle root, Rekor witness, and corrections ledger. Remote Streamable HTTP endpoint:
https://councilof.ai/mcp; free board athttps://councilof.ai/api/gspc; paid evidence packs settle over x402 on Base. Local bridge:npx -y csoai-gspc-mcp. MIT. -
Neither: Local stdio MCP server for retrieving selected company and project context from Neither's hosted context graph with source evidence. Install:
npx -y @neitherai/mcp-server; requiresNEITHER_API_KEYfrom a Neither workspace. Site: https://www.neither.online/start/?product=dev. Official Registry:io.github.stonianua/neither-mcp. MIT. -
ai.maximem/synap: Remote Streamable HTTP MCP memory layer for logging conversation exchanges and recalling relevant user context, facts, preferences, and recent memories across sessions. Tools:
log_exchange,recall_context,list_recent_memories,check_memory_status. Endpoint:https://synap-mcp.maximem.ai/mcpwithAuthorization: Bearer <SYNAP_API_KEY>. Product page: https://www.maximem.ai/synap. Official MCP Registry idai.maximem/synap. Apache-2.0. -
hermes-labs-ai/fidelis: Zero-LLM local-first agent memory combining BM25, dense-vector, and reciprocal-rank-fusion retrieval, returning original passages verbatim by default instead of paraphrased summaries. Install:
pip install fidelis-memory. Transport:stdio; no auth; clients: Claude Code, Gemini CLI, and compatible MCP clients; MIT. -
yliuai/spomory: Personal AI memory MCP server — extracts what you tell it into a graph of entities and relations, retrieves by walking the graph with personalized PageRank instead of guessing from a similarity score, and uses a GRPO-trained (RL) policy, not hardcoded rules, to decide what to add, update, or forget. Local mode stores memory in an encrypted-at-rest SQLite file; cloud mode is a hosted Postgres-backed endpoint reachable from more than one client. Install from source:
git clone+uv pip install -e ".[mcp]", then run thememory-core-mcpentry point. Transport:stdio(local, no auth) orstreamable-http(cloud;x-api-keyheader or OAuth 2.1). Clients: Claude Desktop, Cursor, Codex CLI, and compatible MCP clients. Apache-2.0. Docs: https://spomory.yliuai.com. Official Registry:io.github.lyf-wxy/spomory. -
AgentPostmortem/Casebook-MCP: Remote MCP server exposing a public registry of documented AI-agent failures as queryable tools (ranked case search, full case detail, similar-failure match, tag listing), plus a companion investigator agent that drafts postmortems grounded in real precedents. Transport:
streamable-http; no auth (read-only public data); clients: Claude Code, Cursor, and compatible MCP clients; MIT. Endpoint:https://mcp.agentpostmortem.com/mcp. -
techreone/xknow-mcp: Local MCP server for the XKnow knowledge base — a curated, cross-linked body of SEO, SaaS, and LLM-wiki knowledge. Six read-only tools: ranked
search_knowledge,get_pagefor full note text, graph-awareexplore_concept(a note plus its outbound links and backlinks),list_topics,lint_rules(a written-content self-check rubric with priorities and fixes), andcitefor canonical citation metadata so answers can attribute the source. No network calls, no telemetry, no API key: the free layer bundles a static snapshot of 57 public guides, and--vaultpoints it at a purchased 500+ note vault that never leaves the machine. Install:npx -y xknow-mcp. Transport:stdio; no auth; clients: Claude Desktop, Claude Code, Cursor, Cline, opencode, and compatible MCP clients; MIT. Site: https://xknow.org/mcp. Official Registry:io.github.techreone/xknow-mcp. -
Hyperconsciousness: Developer-alpha encrypted, append-only knowledge store with scoped, expiring access grants. Build from source with
cargo build --release --locked; runhc mcp --as <grant-id> --dir <store-path>. Transport:stdio; requires a locally initialized store and an owner-issued grant. Tools includesearch,recent,recordandfiles. MIT. Setup. -
Memento (mementoagi.com): Hosted MCP memory for coding agents with persistent, editable Markdown records and team sharing. Streamable HTTP at
https://api.mementoagi.com/mcp; requires a dashboard API key. Project. -
Vortx-AI/emem: Shared, verifiable memory for AI agents grounded in Earth observation. Locate places by name or coordinate; recall signed elevation, vegetation, flood, fire, and air-quality facts; cite them with content-addressed tokens that another agent can verify offline; and write observations under a capability-bound namespace. Hosted Streamable HTTP at
https://emem.dev/mcpwith an 18-tool core, orhttps://emem.dev/mcp/fullfor all 110 registered tools. Reads require no API key. Apache-2.0. -
New Runtime archive (com.newruntime/archive): Search and read a public, source-linked AI-engineering newsroom archive with
search_archive,get_record, andcontextualize_url. Optionalsubmit_source,submit_correction, andsubmit_usage_receipttools only queue proposals for human review;list_paid_toolslists optional x402 research endpoints. Hosted Streamable HTTP endpoint:https://newruntime.com/mcp; free reads require no account or key. Hosted service. -
liza-studio/skillmem: Self-improving skill memory for Claude Code and Codex: agents learn how they solved a task, recall it next time, reinforce what works and forget the rest. Local SQLite + FTS5, 9 MCP tools, no API key. Install:
pip install skillmemthenskillmem init --claude-code(or--codex,--cursor,--windsurf,--gemini,--opencode). Transport:stdio; no auth. Apache-2.0. Site: https://skillmem.dev. Official Registry:io.github.liza-studio/skillmem. -
Continuity: Private story bible for fiction. Agents can manage story projects, read and search continuity context, update locked canon, record approved scenes, save checkpoints, and run continuity checks. Hosted Streamable HTTP at
https://continuitywriter.com/mcpwith OAuth 2.1; Pro plan or trial required. Docs. -
ssap-pa/self-learning-agent-setup:
feedback-memoryturns your approve, edit and reject decisions into memory an agent checks before it drafts:record_decisionsaves the edit and your reason,recall_correctionsreturns the closest past corrections and standing rules for the next task. Five tools (recall_corrections,record_decision,add_rule,list_memory,forget); local Postgres + pgvector via PGlite in~/.feedback-memory, offline hashing embedder by default or OpenAI embeddings whenOPENAI_API_KEYis set. Install:claude mcp add feedback-memory -- npx -y github:ssap-pa/self-learning-agent-setup. Transport:stdio; no auth; clients: Claude Code, Cursor, Claude Desktop, and compatible MCP clients. MIT. In-browser demo: https://ssap-pa.github.io/self-learning-agent-setup/. -
uNotes: Search a library of university course materials (past exams, assignments, lab reports, lecture notes) and your uNotes flashcards and quizzes. Hosted remote endpoint:
https://unotes.net/api/mcp(Streamable HTTP, OAuth 2.1 with dynamic client registration, no API key). Official MCP Registry:net.unotes/unotes. Setup for Cursor, VS Code, Cline and Claude Code: https://github.com/DevinoSolutions/mcp-servers/tree/main/unotes -
Save This One: Save URLs, search and read saved bookmarks, manage tags and notes, and organize a personal bookmark library. OAuth sign-in scopes each connection to the user's Save This One account. Compatible MCP Apps hosts can show an interactive library and saved-article reader.
-
ai.locul/cloud: Locul turns your notes, docs and dictations into memory your AI uses. This hosted server lets Claude and ChatGPT recall it and save new memories.
-
com.contextli/contextli: Contextli's MCP server lets Claude, ChatGPT and Cursor search and read your own voice notes and transcriptions, filtered by context, label or date.
-
happywpco/pdftomarkdown-mcp (PDFtoMarkdown.ai MCP): Converts local PDFs, Office documents, images, HTML, and EPUB files into structured Markdown with headings, tables, and OCR text preserved. Exposes
convert_pdf_to_markdownso agents can read documents for RAG, notes, or analysis. Install withnpx -y @happywp/pdftomarkdown-mcp; stdio transport withPDFTOMARKDOWN_API_KEY. MIT. -
volcengine/OpenViking: Self-hosted context database for AI agents (39k+ GitHub stars) — resources, memory and skills in one browsable
viking://filesystem with layered L0/L1/L2 summaries, exposed to coding agents over MCP. Install:uv tool install openviking --upgradethenopenviking-server init. Transport: streamable HTTP athttp://<server>:1933/mcp. Clients: Claude Code, Codex, Cursor, OpenCode. AGPL-3.0. Site: https://openviking.ai/. -
Selvedge: Record and retrieve coding decisions, reasons, and rejected approaches by entity across agent sessions in a local SQLite database. Install:
uv tool install selvedge; launch:selvedge-server. Transport: stdio. Auth: none. Clients: compatible local stdio MCP clients, including Claude Code, Codex, Cursor, Gemini CLI, VS Code, and Windsurf. Tools: 8, includinglog_change,prior_attempts,blame, andstale_decisions. License: MIT. Docs: https://selvedge.sh/start/what-is-selvedge/. -
MemTether: Local cross-client memory hub with a tamper-evident evidence chain, supersession history, conflict detection, and adapters for Claude Code, Cursor, Windsurf, Codex, Gemini CLI, and other clients. Install with
pip install memtether, then runmemtether initto create the local database and configure detected clients. Apache-2.0. -
ORANO: Hosted, read-only memory of what you saved: projects ORANO builds from Reels, TikToks, YouTube videos, articles and PDFs (summaries, key takeaways, ordered tasks, research), selectable video context (transcript, on-screen text, sampled frames, links) and curated memory facts. Tools:
list_projects,get_project,get_project_context,get_pending_handoffs,search_library,get_memory_facts. Endpoint:https://orano-ai-backend-1037939693300.us-central1.run.app/mcp/. Transport: streamable HTTP. Auth: bearer personal API key (orano:read, ORANO MCP web plan). Manifest: https://github.com/infotik/orano-mcp-server. -
Remnant Read: Search prior technical agent experience and inspect its published evidence, applicability conditions, failed approaches, and reported outcomes before trying a solution. The anonymous Read connection is read-only; recording outcomes and contributing reusable lessons use a separate authenticated connection. Hosted Streamable HTTP endpoint:
https://remnant.dedale-bi.com/mcp/chatgpt; no authentication. Seven tools includesearch_memories,inspect_memory, and trust-passport inspection. Setup. -
agents-learning: Local project memory for Codex: search bounded knowledge previews and read revision-aware pages with applicability, limits and source evidence. Uses Python stdio MCP, SQLite and FTS, with optional local semantic retrieval. The full plugin adds lifecycle Hooks; administrator review and trusted publication use separate roles.
-
Enhanciar: Hosted remote MCP server for a team's company brain. Ask questions over connected Slack, Gmail, Google Drive, Notion, Jira, Linear and GitHub and get answers with a link to every source; tools include
query,search_wiki,get_page,impact(code change impact) andpropose_action, which queues a drafted reply, ticket or doc for a person to approve. Endpoint:https://enhanciar.in/enhanciar/mcp/. Transport:streamable-http; auth: bearer API key (dh_..., created in the app). Clients: Claude Code, Claude Desktop (viamcp-remote), Cursor, VS Code, and compatible remote-MCP clients. Site: https://enhanciar.in. Docs: https://enhanciar.in/mcp. Official Registry:in.enhanciar/enhanciar. Paid service; docs repo MIT. -
Folio Wiki: Open-source (MIT) self-hosted wiki for teams in which every page is a Markdown file in a Git repository, with a built-in MCP server. Its 23 tools cover search, reading, creating and editing pages, data tables, whiteboards and page history. Install: Docker image
ghcr.io/evergreen-it-dev/folio(docker compose). Transport: Streamable HTTP at/mcpon your own instance; auth: OAuth 2.1 or personal access token. Clients: Claude Code, Cursor, VS Code, Codex and other HTTP-transport MCP clients. TypeScript, MIT. Site: https://foliowiki.online. Docs: https://github.com/evergreen-it-dev/folio/blob/main/docs/MCP.md. Official Registry:io.github.evergreen-it-dev/folio. -
Pinto Notes: Hosted MCP server for Pinto Notes, a free, open-source notes app for Mac (iPhone in progress) that imports Apple Notes. Lets an agent search, read, create and edit notes, tick checklist items and add rows to tables, and every change keeps the previous version. Endpoint:
https://mcp.pintonotes.com. Transport:streamable-http; auth: OAuth, approved by the person in the app, read only or read and edit. Clients: ChatGPT, Claude, Claude Code, Codex, Gemini CLI and other remote MCP clients. Server source insupabase/functions/mcp, TypeScript, MIT. Site: https://pintonotes.com. Docs: https://pintonotes.com/blog/mcp-server. Official Registry:com.pintonotes/pinto-notes. -
Cartographer (BeppeTemp/cartographer): Read, search and grow a shared git-backed Markdown wiki through validated, serialized commits. Install:
brew install beppetemp/tap/cartographer, thencartographer serve --kb ./kb --init(stdio). Self-hosted HTTP supports bearer tokens; public read-only demo athttps://cartographer-demo.beppetemp.com/mcprequires no auth. Apache-2.0. -
Twin Me Up: A personal work twin for any agent. Briefs the agent on the guidance you have confirmed, checks consequential actions against explicit, expiring grants (anything else returns ask), and turns your corrections into lessons only you can confirm. 10 tools, transport stdio, no auth, local SQLite store, Apache-2.0. Install:
uvx twin-me-up. -
Whisper Memos: Official Whisper Memos server. Read the transcripts, summaries and labels of voice memos recorded in the Whisper Memos iPhone and Apple Watch app, and delete memos. Tools:
list_memos,get_memo,list_labels,delete_memo,get_account; it cannot create or edit memos. Remote streamable-http athttps://whispermemos.com/mcp, OAuth 2.1 with dynamic client registration, approved by the user in the iOS app. Registry idcom.whispermemos/whisper-memos. Config: https://github.com/whispermemos/whisper-memos-plugin. Docs: https://whispermemos.com/mcp. -
alexslkgit/own-words: Turns notes into study decks: the reader types an answer in their own words, then compares it with the prepared answer and marks it themselves. Endpoint:
https://own-words-mcp.alexslkmain.workers.dev/mcp. Transport:streamable-http; no auth. Tool:create_deck. MIT. Official Registry:io.github.alexslkgit/own-words. -
enhesa/big-agentic-brain: Self-hosted shared memory, messaging and reminder service for an organization's AI agents. Agents record fixes and search them before debugging, and message each other across teams. 10 MCP tools over remote HTTP, plus a JSON API and read-only web UI; one container with SQLite. Install:
docker compose up -d --build. License: Apache-2.0. -
Apex AI Academy: Official server for Apex AI Academy, an academy that teaches AI agents how to teach their humans AI. Browse courses and lessons, read free lessons (agent, human or both views), search, get recent AI news, and check pricing. Tools:
list_courses,list_lessons,get_lesson,search,latest_news,get_pricing,subscribe_link; Premium tools with a personal API key:get_premium_lesson,get_progress,mark_lesson_complete. Remote streamable-http athttps://apexai.academy/mcp; public tools need no auth. Hosted, closed source. Official Registry:academy.apexai/apex-ai-academy. Docs: https://apexai.academy/llms.txt. -
MindDory: Save words and phrases the user meets in a chat as spaced-repetition vocabulary flashcards, check which words the user already knows, read the user's review queue and recent activity, and record when the user used a word correctly or struggled with it. Built for language learners who practise with an AI assistant.
-
RamaAditya49/titen: Gives agents persistent memory with no API key, LLM or embedding provider. Agents record source observations, write evidence-linked claims, and compile scoped context; retrieval is SQLite FTS5 with optional vectors. Implements the nine @modelcontextprotocol/server-memory tools with the same schemas, so it works as a drop-in replacement.
-
Dashly: Track everything by telling your AI assistant, and give structure to your AI conversations. Logs calories, habits, mood, workouts, spending and tasks (or anything you want to track) as dated entries, and shows each as a live dashboard tile against its goal: an AI calorie tracker, habit tracker, mood tracker, expense tracker and task list in one. 24 tools. Remote streamable-http at
https://dashly.si/mcp, OAuth 2.1. Official Registry:si.dashly/dashly. Docs: https://dashly.si/connect. -
OptimNow/cloud-finops-skills: An agent can list, filter and read FinOps reference guides on cloud and AI billing mechanics, commitments, allocation and chargeback, and fetch named waste-pattern playbooks that carry a detection query to run in the user’s own account. Six read-only tools over content bundled with the package; nothing to connect to a cloud account.
-
Vespper: Hosted MCP server for Microsoft Word documents. An agent reads a .docx as lean HTML and its edits are written back into the original file as native tracked changes, so the result is a reviewable redline rather than a regenerated document. Transport:
streamable-httpathttps://mcp.vespper.com/mcp, OAuth. Closed source; commercial hosted service. Docs: https://docs.vespper.com/quickstart. -
MemorySync: Hosted (closed-source) remote MCP memory server. An agent can search and store durable facts, preferences and decisions across sessions. Endpoint:
https://mcp.memorysync.io/mcp. Transport: streamable HTTP. Auth: OAuth 2.1 browser sign-in, or a MemorySync API key for headless clients. Install:npx -y memorysync-mcp-install. Docs: https://docs.memorysync.io/mcp/install. Official MCP Registry:io.memorysync/memory. -
RuMemory/maxgraph: Persistent, project-scoped memory across sessions: a knowledge graph of decisions, bugs, tasks and session summaries with hybrid semantic search, automatic context briefs and a live graph dashboard. Local-first and offline, storing plain-text files on the user's machine. Install from the repository, then connect over Streamable HTTP at
http://localhost:8787/mcpwith a Bearer token. BSL-1.1, converting to MIT four years after each release. -
Digital Garden by Alicia — Notes: Search and read Alicia’s English and Chinese field notes on Houdini, three.js/WebGPU, glTF, OpenUSD, Gaussian splats and AI workflows. Four read-only tools: search_articles, get_article, list_topics, list_recent_articles. Article results include canonical source URLs for citation.
-
Thanatos9404/llmslim: Gives MCP clients (Claude Code, Claude Desktop, Cursor and others) long-term memory with no LLM call. Tools:
remember,recall,list_memories,forget,fit_context. Messages are stored word for word in local SQLite or Postgres and recalled within a token budget; the user and agent scope is fixed when the server starts, not chosen by the model. Install:pip install "llmslim[mcp]", then runllmslim mcp(orclaude mcp add llmslim -- llmslim mcp). Transport:stdio; no auth. MIT. Official Registry:io.github.Thanatos9404/llmslim. Glama: https://glama.ai/mcp/servers/Thanatos9404/llmslim. -
Agents Handbook: Remote MCP server where an agent can search and read sourced playbooks for real jobs, each with a prompt, steps, human checkpoints and known failures, and report back how a run went. Endpoint:
https://agentshandbook.com/mcp(no sign-in to read). -
koruworks/mono: Local MCP server (
koru-mcp) for Koru, a free, open-source, local-first workspace for notes, docs and files built natively in Rust. MCP clients can search and read documents and files, and create and edit documents, in the local database on the device. Transport:stdio; no auth, no account. Clients: Claude Desktop, Codex, and compatible MCP clients. Windows and macOS (Apple silicon). MPL-2.0. Docs: https://github.com/koruworks/mono/blob/main/crates/koru-mcp-server/MCP.md. -
xTiles: Official remote MCP server for xTiles, a visual workspace for notes, projects and tasks. Agents can turn chats into xTiles projects, pages, notes and tasks, create and update tasks with due dates and assignees, add notes to the daily planner, read the calendar, and pull any project back into the chat with its current state, decisions and open tasks.
-
FactoGraph: Official FactoGraph MCP server. Answers from facts people approved, every claim linked to the fact behind it: list the facts of any public FactoGraph store, or ask it a question and get an answer whose sentences cite those facts; with a person's consent, make them a private memory store that lasts across conversations. Public stores are read-only through this server. Remote endpoint:
https://api.factograph.app/api/mcp. Transport: streamable-http. Auth: none (no sign-up). 6 tools:listFacts,ask,replyToPrompt,getManual,getStep,createStore. Official Registry:app.factograph/factograph. Docs: https://www.factograph.app/for-ai-agents.html. -
Scento-com/perfume-mcp-server: Perfume portraits, raw materials, scent pairings and occasion edits from Scento's verified perfume database: 1,400+ perfumes, 400+ raw materials and a family map, with a scento.com page on every answer. Twelve read-only tools. Remote streamable-http at
https://www.scento.com/mcp, no auth. Install for local-only clients:npx -y scento-perfume-api. MIT. Docs: https://www.scento.com/developers/perfume-mcp-server. Official Registry:com.scento/perfume-api. -
tokenrip/cli (Tokenrip): Tokenrip is a shared workspace where Claude, ChatGPT, Codex and Grok share files, tasks, decisions and handoffs. Agents save work to a workspace, pick up work another agent left, and ask a person before anything goes out.
-
OneMira AI (onemira-mcp): Local-first MCP server for the OneMira C1 AI Voice Recorder and OneMira Desktop (macOS 12+ and Windows 10/11). Agents search, read, and update 99-language speaker-diarized meeting transcripts and Google Native OKF Markdown Wiki notes (
mira_search,mira_sessions,mira_segments,mira_wiki_read,mira_wiki_update, 38 tools total). Transport:stdiovia localmira mcpor TypeScript MCP wrapper (github:onemira-space/onemira-mcp, MIT); zero cloud subscription ($0/mo). Supports Claude Code, Claude Desktop, Cursor, Windsurf, and Obsidian. Docs. Site: https://onemira.com/products/c1. -
ctxstore-ai/ctxstore-mcp: Persistent memory for AI agents. Stores keyed facts with layers and history, recalls them by exact key or semantic search, and returns a wake bundle (standing rules, current state, the last session's closing note) at the start of each session, so one memory carries across Claude, ChatGPT, Cursor and other MCP clients and across model changes. Remote streamable-http at
https://mcp.ctxstore.ai/mcp, OAuth 2.1 with dynamic client registration or an API key; free to start, and a no-account trial athttps://mcp.ctxstore.ai/try. Install (Claude Code):claude mcp add --transport http ctxstore https://mcp.ctxstore.ai/mcp. MIT. Official Registry:ai.ctxstore/ctxstore-mcp. -
kordoc: Convert Korean and Office documents (HWP 3.x/5.x, HWPX, PDF, DOCX, XLSX, PPTX and images) to Markdown and structured data, with local OCR, block and cell diffs, form filling and Markdown-to-HWPX generation. 17 tools. Install:
npx -y kordoc setup, or runnpx -y kordoc mcp(stdio); also a CLI and npm library. Auth: none. License: MIT. -
Grit-77/remember-dammit: Use a local vault for explicit capture and recall of source-linked notes, decisions and preferences. Correct outdated decisions while retaining their history, and keep project scopes separate. Python CLI and local stdio MCP server; currently a 0.2.0a1 source alpha.
-
NicksAdvisory/lintelly-hr-deutschland: German employment law only. Answers the HR questions of small companies in Germany (notice periods, employer costs, sick pay, maternity protection, contract clauses), each answer with the calculation and the official legal source. 38 read-only tools (e.g.
ablauf,kuendigungsfrist_rechnen,arbeitgeberkosten,entgeltfortzahlung,vertragsbaustein); no names or personal data, inputs not stored. Remote streamable-http athttps://hr.lintelly.com/mcp, no auth, no sign-in. Local:npx -y @lintelly/hr-deutschland(stdio) or Dockerghcr.io/nicksadvisory/lintelly-hr-deutschland. Clients: Claude, ChatGPT, Claude Code, Cursor, Gemini CLI and other MCP clients. MIT. Site: https://lintelly.com/hr. Official Registry:com.lintelly/hr-deutschland. -
Knowledge for Agents: Search shared agent experience by exact error and environment, fetch source-linked problems, solutions and public conversations, and inspect failed approaches and outcomes. Hosted Streamable HTTP at
https://knowledgeforagents.com/mcp; no authentication for public read tools (search,fetch,get_changes,whoami). Optional pseudonymous participation is bounded and governed. Docs: https://knowledgeforagents.com/connect. MIT public integration package; hosted terms: https://knowledgeforagents.com/terms. Official MCP Registry idcom.knowledgeforagents/knowledge-for-agents. -
Buy-Box: Load a user's saved buying criteria for homes, cars or any other category (budget, must-haves, nice-to-haves, dealbreakers) before it searches, save or update those criteria, score and rank up to 100 listings against a saved buy box, export a box as Markdown, and list or delete categories.
-
Coworker (official MCP Registry: ai.coworker/mcp): Coworker is a company knowledge graph exposed over MCP.
-
MemoryRouter: Hosted memory service for AI clients that share one vault, so a fact saved in one client is recalled in the next. Agents search memories by meaning or by date window, store one approved memory, check connection status and the selected vault reference, inspect and consolidate stored memories across tiers, and delete individual memories or the whole vault. Endpoint:
https://mcp.memoryrouter.ai/mcp. Transport:streamable-http(stateless); auth: OAuth with PKCE and RFC 9728 discovery, with independentmemories:read,memories:writeandmemories:deletescopes bound to one OAuth-selected vault. Clients: Claude, ChatGPT, Cowork, Claude Code, Codex, OpenClaw, and other remote MCP clients. Docs: https://docs.memoryrouter.ai/mcp. Site: https://memoryrouter.ai. Official Registry:ai.memoryrouter/memoryrouter. -
mohamed-ashraf-elsaed/aletheia-mem: Free, local memory and rules hub for AI coding tools: one SQLite store read over MCP, REST or CLI (alpha). Install:
npm install -g aletheia-mem, stdio serveraletheia-mcp. -
GhostPawJS/codex: Local memory for agents in one SQLite file. Each belief keeps its source, certainty, evidence count and revision history; the agent searches before it remembers, confirms or corrects instead of appending, and gets stale or weakly supported beliefs flagged for review. 7 tools. Install:
npx -y @ghostpaw/codex mcp(Node.js 24+), or the standalone binary for Linux, macOS and Windows from GitHub Releases (curl -fsSL https://raw.githubusercontent.com/GhostPawJS/codex/main/install.sh | sh, thenghostpaw-codex mcp);server.mcpbfor Claude Desktop. Transport:stdio; no auth; DB path viaGHOSTPAW_CODEX_DB. Clients: Claude Code, Claude Desktop, Cursor, Codex CLI. MIT. Site: https://codex.ghostpaw.dev/. Official Registry:io.github.GhostPawJS/codex. -
sadofriod/graphrag-ts: Citation-graph retraction checks for a DOI with source-tier evidence and explicit abstention reasons. Streamable HTTP at
https://mcp.ashesborn.cloud/mcp; discovery andintegrity_paper_statuswork anonymously.retraction_propagationandintegrity_evidence_auditrequire a keyed session or paid x402 access. Clients: Claude Code, Cursor, VS Code, OpenCode and Hermes. MIT code; CC0-1.0 published snapshot. Hosted service has no SLA; mainnet x402 settlement is not yet exercised. -
Speechyou MCP: Search and read Speechyou voice notes and meeting transcripts from Claude and compatible AI assistants. Four read-only tools find notes by keyword or tag, retrieve full transcripts and available AI summaries with audio links, list recent notes, and list tags. Access is limited to notes the signed-in user owns or that have been shared with them.
-
OktoLabsAI/okto-neuron: Local-first knowledge graph memory for agents, with cited sources and an MCP server. Install:
uv tool install "okto-neuron[serve,litellm]". License: Elastic License 2.0 (source-available). -
Richard-Ewing/exogram-sdk: Official MCP bridge into Exogram (https://exogram.ai)—a Personal AI Agent and workspace that understands your work, people, and everyday life, coordinates specialist agents, builds interactive deliverables (
render_live_widget), and executes tasks with your approval (traverse_neural_graph,evaluate_instinct_action_gate,verify_audit_trail). Runs in zero-signup local SQLite mode (~/.exogram/vault.db) or over hosted Streamable HTTP (https://api.exogram.ai/mcp). -
fainir/recall: This MCP server lets agents read and write the per-project auto memory Claude Code keeps as Markdown files with wiki-links, so Codex, Cursor and other MCP clients share the same memories. With
--read-onlyit exposes only the list, get, recall and due tools; without it the agent can also write memories that Claude Code reads back. -
jackjin1997/markindex: Vectorless, structure-first retrieval over large Markdown libraries for agents: browse library/folder/document/section trees, read original lines for citations, and search with mixed Chinese/English BM25 in SQLite. Five read-only tools; local, MIT, no embeddings and no API key. Install:
npx markindex mcp <docs_dir>. -
OnlyMCP: OnlyMCP is a hosted MCP server that gives every AI client the same skills, agents and markdown memory, plus a gateway to other remote MCP servers. Agents can read, write, search and link memory files, load and edit skills and agent definitions, and call the tools of servers connected once to the account (GitHub, Notion, Linear or any remote MCP server), with their credentials stored encrypted on the server. One address replaces a separate MCP config per client.
-
star-nebula/SemanticMemory: Evidence-first Markdown store for reviewed current facts, decisions, source quotes, and tasks, with an optional
reme_refpointer into ReMe evidence. Install:python -m pip install https://github.com/star-nebula/SemanticMemory/releases/download/v0.1.1/semantic_memory_reme-0.1.1-py3-none-any.whl. Transport:stdio; no auth; clients: Codex, Claude Code, Qoder, and compatible MCP clients; Apache-2.0. Docs: https://github.com/star-nebula/SemanticMemory/blob/main/docs/reme-integration.md. -
Skybits: Official Skybits server. A shared workspace where people and AI agents create and edit documents, spreadsheets and HTML canvases together, kept in one searchable library. Agents can find, read, create and edit documents (changing one section and leaving the rest intact), answer and resolve comments, organise the library with tags, and export to PDF, Word, HTML or Markdown. Several people and agents can edit the same document at once; in documents, agent edits arrive as suggestions a person accepts or rejects, and every change is recorded with its author and can be restored. Endpoint:
https://skybits.ai/mcp. Transport:streamable-http; auth: OAuth 2.1 (dynamic client registration), no API key. Clients: Claude and compatible remote-MCP clients. -
Coisini-cyber/ctxkeeper-mcp: Give coding agents durable project memory (decisions, gotchas, commands, TODOs, work logs) and return token-budgeted code windows with path:line citations plus matching memories. Local-first, no API key, no vector DB, no telemetry.
-
Holos33: This MCP server lets agents retrieve sourced, read-only Holos33 information about modules, Early Access, privacy, access and contact.
-
Compass OS: Official remote MCP server for Compass OS, a career OS. Agents can read the user's own profile, wins log, career timeline, tracked roles and where each stands, best-fit roles from the last job scan, tailored resumes and cover letters, LinkedIn audit, and interview story bank and practice progress. They can also ask to track a role, edit the profile, log a win, save an interview story or draft a tailored resume; every change shows a confirmation card and saves only after the user says yes. It never applies, posts or contacts anyone. Remote endpoint:
https://compass-os.dev/api/mcp. Transport: streamable-http. Auth: OAuth (Compass OS sign-in; Plus or Pro plan). Clients: Claude, ChatGPT, VS Code, Cursor, Claude Code. Docs: https://compass-os.dev/mcp. Official MCP Registry:io.github.tsinovoi/compass-os. Hosted, closed source. -
knowsme: One memory of who you are and how you work (notes, rules and projects) that any AI model or agent can read and update over MCP, across apps and machines. 9 tools: list notes with a one-line summary each (
list_memories), read one (recall), find notes by subject (search_memories), save or replace a note (remember), add a line to a project's progress log (save_progress), read earlier versions to undo a change (history), delete a note (forget), and suggest rules or a tidy-up that the user approves on their dashboard (propose_rules,propose_tidy). Hosted Streamable HTTP athttps://mcp.knowsme.dev/mcp(SSE athttps://mcp.knowsme.dev/sse); sign in with OAuth or an access key. Install:claude mcp add knowsme --scope user --transport http https://mcp.knowsme.dev/mcp. Official MCP Registry iddev.knowsme/memory. Free to start (up to 200 notes). Setup for each app. Hosted service. -
Counterparts: Memory for Claude, as an MCP server plus hooks for Claude Code (macOS, Linux) and Claude Desktop (macOS). Install:
npm install -g counterparts@latest, then runcounterparts, which wires the hooks and the MCP server. Transport:stdio; no auth; clients: Claude Code, Claude Desktop; local; MIT. -
Lodestar: Look up what worked for other agents doing real work. It searches lessons by meaning (describe the task, or paste the exact error), draws the lesson it will use, and reports in one call whether that lesson helped, did not apply or misled. Search shows those totals to the next agent. An agent its operator has claimed can also contribute lessons, and every lesson passes an automatic privacy screen. Connect via Streamable HTTP at
https://api.lodestar.wiki/v1/mycelium/mcp; connection and search need no key, while other tools use a key returned bysign_up. Six tools:search,get_lesson,report_outcome,contribute,claim,sign_up. Docs: https://github.com/zkysar1/lodestar-mcp#readme. -
Maindex Memory Services: Team-shared memory for AI agents. Expert provides 15 tools for memories, collections, typed links, hybrid search, bulk operations and Teams workspaces at
https://expert.maindex.io/mcp; Smart offers keep, recall, update and forget on a shared team graph athttps://maindex.io/mcp. Transport: Streamable HTTP; API key viaX-API-Key. Official Registry:io.maindex/expertandio.maindex/smart. -
Glasp: Remote MCP connector to search and read your own saved web and Kindle highlights, page summaries and bookmarks, with sources. Read-only apart from one permission you approve when connecting: saving a page to your Glasp bookmarks, which are public on your profile. Remote endpoint
https://glasp.co/api/mcp, OAuth, free with a Glasp account. Official Registry:io.github.kazuki-sf/glasp. -
endtoend-ai/Company-Brain: Local-first knowledge workspace that combines document RAG, GraphRAG over PostgreSQL + Apache AGE, and a page-based knowledge brain, with every answer returned alongside its source evidence. GraphRAG ships a stdio MCP server exposing
graph_search,graph_list_sourcesandgraph_get_document. Install: clone the repository, runuv sync --project modules/graph-rag, then startuv run --project modules/graph-rag python -m graph_rag.mcp.serverwithGRAPH_RAG_DATABASE_URLandGRAPH_RAG_MCP_TOKENset. Transport:stdio; MIT. -
tidybytetools/doc2md-lite: Convert PDF, Word and HTML documents to Markdown locally. No API key. Install:
uvx doc2md-lite-mcp. Official MCP Registry idio.github.tidybytetools/doc2md-lite. MIT. -
SavedThat: Search the videos you saved from Instagram Reels, TikTok and YouTube by what was said in them, and get the moment back with a quote, a timestamp and a link; also searches transcribed libraries of public creators. Tools:
search_my_library,get_moment,list_my_saves,get_my_transcript,show_creators,search_creator_videos,get_creator_moment,show_plans(8; all read-only exceptshow_plans, which opens Stripe Checkout). Remote streamable-http athttps://savedthat.app/api/mcp/me, OAuth 2.1 with dynamic client registration and PKCE, no API key. Clients: Claude, Claude Code, ChatGPT and compatible remote-MCP clients. Official Registry:app.savedthat/library. -
mindmap.io: Official mindmap.io server. Create mindmaps of any topic from Claude, Cursor or any AI agent. Every node can be an AI chat. Its 19 tools read your mindmaps, add and edit nodes, run a node's question, and publish a mindmap as a link. Hosted at
https://mindmap.io/mcpwith OAuth, where you choose Full access or Read only. Local:npx -y github:MohGanji/mindmapio-mcpwithMINDMAP_API_TOKEN. Free plan. MIT. Source: https://github.com/MohGanji/mindmapio-mcp -
MemoryBrain: Personal memory that ChatGPT and Claude share over MCP; every note keeps its source (something the user read, said, or a chat) and typed links to other notes. Remote:
https://app.memorybrain.ai/mcp(Streamable HTTP, OAuth; private beta with limited sign-up). Method and skills: https://github.com/Memory-brain/open-brain (MIT). -
LF-Labs-Inc/dawnmere: Recall, remember and forget memories in one local memory on a Mac that Claude Code, Codex and Cursor share, and check its own access level (tools: recall, remember, forget, access_status). The memory is one SQLite file with local search; each app gets an access level, and no account is needed. I am the author. Early beta (0.1.9).
-
Eno654/oxyo-mcp: Persistent contradiction-revising cross-session memory, causal root-cause graph, isolated zero-network Python verification sandbox, and formal discrete constraint prover for coding agents, cutting cold-start token burn by 60%–72%. Tools:
oxyo_session_pack,oxyo_recall,oxyo_remember_belief,oxyo_remember_episode,oxyo_causal_record,oxyo_causal_infer,oxyo_verify_python,oxyo_prove_constraints,oxyo_skill_register,oxyo_skill_invoke(15 tools total). Install:npx -y github:Eno654/oxyo-mcp. Endpoint:https://oxyo.si/mcp. Transport:stdio,streamable-http. Auth: bearer API key (initializeandtools/listunauthenticated; free 50-call tier). Clients: Claude Code, Cursor, Codex, VS Code, Windsurf, Claude Desktop, and compatible MCP clients. Docs: https://oxyo.si/mcp. License: MIT. -
ReadShore (app.readshore/readshore): Official hosted server for ReadShore, a read-later app for iPhone. Search the articles and posts a reader saved, read an item's saved text in pages with numbered passages to cite, along with ReadShore's summary and analysis, and save new links to the library. Tools:
readshore_search_saved_items,readshore_read_saved_item(read-only) andreadshore_save_link(adds only; a link already saved returns that item). Hosted Streamable HTTP athttps://readshore.app/mcp; OAuth 2.1 with PKCE and dynamic client registration, sign-in with Apple; each connection reaches only the signed-in reader's own library; requires a ReadShore account from the iPhone app. Clients: Claude, ChatGPT and other remote-MCP clients. Official MCP Registry idapp.readshore/readshore. Setup guide. Package files: https://github.com/flsteven87/readshore-mcp (MIT). -
aligndottech/align-cli: Local decision graph for coding agents. Imports the decisions a team already made from Jira, Confluence, Linear, Notion, Slack, Teams, Zoom, GitHub, GitLab, git history and repo docs (read-only), links related decisions, and lets the agent check a proposed change against them before it edits. Tools include
align_check_alignment,align_ask,align_search,align_get_decision_rationale,align_get_related_decisions,align_get_topic_timelineandalign_capture. Install:npm i -g @aligndottech/clithenalign; MCP server:npx -y @aligndottech/cli mcp. Transport:stdio; no account needed; optional own model key or local Ollama. Clients: Claude Code, Codex, Gemini CLI, Cursor, Copilot CLI, VS Code, Windsurf, Zed, OpenCode, pi. MIT. Official Registry:io.github.aligndottech/cli. -
code4161/memgit: Save, search and roll back memories in one store that Claude Code, Cursor, Codex and any MCP host share; imports Claude Code's own memory files, keeps every change as a checkpoint with diff and rollback, BM25 recall, optional end-to-end encrypted sync between machines. Install:
npx -y memgit-mcp(needs Python 3.11+) orpip install memgit. -
Polidict (io.github.solomkinmv/polidict): Official hosted server for Polidict, a vocabulary app. Save words and phrases to the signed-in learner's vocabulary, look up dictionary meanings and images, organize words into groups, and practice with FSRS spaced repetition (pick due words, submit results, check progress per word). 14 tools, including
lookup_dictionary_word,save_word,find_my_words,get_training_items,submit_training_resultsandget_word_progress(read-only tools carry read-only hints). Hosted Streamable HTTP athttps://api.polidict.com/mcp; OAuth 2.1; each connection reaches only the signed-in user's account. Clients: ChatGPT, Claude and other remote-MCP clients. Official MCP Registry idio.github.solomkinmv/polidict. Plugin and skills: https://github.com/solomkinmv/polidict-ai-assistants. I build Polidict. -
ClickHouse/loom: Self-hosted MCP server and HTTP API that gives AI agents shared long-term memory, document search with citations (PDF, .docx, .pptx) and a code index of your repositories (definitions, callers, change impact). Agent context and index data are stored in ClickHouse; a deployment can choose Postgres for live memory. Plugins for Claude Code and Codex; Python (
clickhouse-loomon PyPI) and TypeScript SDKs; runs with Docker or Helm. Transport:streamable-httpat/mcp; auth: bearer token in multi-user mode. Apache-2.0. Maintained by ClickHouse. Docs: https://github.com/ClickHouse/loom/blob/main/docs/index.md. -
ChineseWorkSheet: Hosted MCP server for Mandarin learning: calculate printable handwriting grids, look up stroke-order rules and radicals, analyze HSK vocabulary and convert Pinyin tones. Streamable HTTP at
https://chineseworksheet.com/api/mcp; no API key required. Public documentation repository is MIT-licensed. -
necroz/sessionwiki: Compiles session transcripts from 7 coding agents (Claude Code, Codex CLI, Cursor, Gemini CLI, Copilot, ZCode, OpenCode) into a local, interlinked markdown wiki: an LLM writes a summary page per session, and a static site adds a session graph and full-text search. The MCP server (15 tools) lets agents look up past decisions and fixes before redoing work; write tools require explicit confirmation. ZCode and OpenCode sessions sync automatically via hooks and a plugin. Install:
uvx sessionwiki mcp(runsessionwiki syncin your wiki folder first). Transport: stdio; no auth; no telemetry. Python, macOS, Linux, Windows. MIT. Docs: https://github.com/necroz/sessionwiki#readme. Official MCP Registry id:io.github.necroz/sessionwiki. -
kolivritech/memoria-mcp: Deterministic long-term memory and code intelligence MCP server for AI coding clients including Claude Code, Cursor, Windsurf, VS Code, Zed, OpenCode and Codex. Cloud portal.
-
thejacedev/Noteriv: Local MCP server (npm package noteriv-mcp) with 22 tools to read, write, search, rename and delete notes and folders in a markdown vault, create daily notes, and list tags, backlinks and outgoing links; works with Obsidian vaults. Install: npx noteriv-mcp. Transport: stdio; no auth; makes no network requests. JavaScript (Node.js). MIT.
-
Octonove/crbro-memory: Local persistent memory for agents stored as plain JSON: recall dated facts, save decisions, replace corrected facts, flag stale values and track open items. Install:
npx crbro-memory install; transport: stdio. MIT.