Zeeq ("zeek") is an MCP server that provides teams with visibility into which content is being activated and used in agent context.
This telemetry lets teams see how agents are using context to affect code authoring.
The same context origin and telemetry are applied to code reviews which can be invoked in the agent loop (versus in CI on the PR).
It is designed to be self hostable, relatively consolidated on a few technologies (can run entirely on Postgres; no additional infrastructure), yet operationally scalable.
Zeeq was originally built as an internal platform at Motion ($500m, series C, YC startup) to shape agentic coding output and is now released as an open source tool.
It is a key part of Motion's internal agentic coding platform, which is used by a team of 40+ engineers to ship with high confidence using AI coding tools.
Zeeq is designed to be easy to setup and run using Mise to install the necessary runtime components.
The backend and Aspire host use .NET 11. Mise installs the SDK version pinned in .config/mise.toml; global.json selects the same SDK for CLI commands.
For local setup see: https://zeeq.ai/docs/managing-zeeq/local-development
Start the local stack with mise run up (GCP Pub/Sub) or mise run up aws (Floci AWS emulator). Stop it with mise run down.
For Claude Code telemetry, keep ZEEQ_OTEL_ACCESS_TOKEN in your private environment and start the client with mise exec -- claude. See the telemetry setup and recovery guide. The MCP login and telemetry token are separate.
For local sign-in without external provider credentials, use the 0factor development users.
Deploying to Google Cloud? See: https://zeeq.ai/docs/configuration/gcp-runtime
Zeeq AI Copyright (C) 2026 Zeeq Labs, Inc
Zeeq is licensed under the GNU Affero General Public License, version 3 or later. See LICENSE for details.