聖地巡礼 Animichi

AI-powered pilgrimage search and route planning for anime sacred sites

Try it live | Architecture | Deployment

English | 日本語 | 中文

---

Tell the agent an anime title or a location in natural language. It finds real-world pilgrimage spots, shows them on a map, and plans a walking route — all in one conversational turn.

How It Works

User text  →  native Pi agent (workers/edge/src/agent/)
                 ├── resolve_anime  → catalog Worker title resolve; Bangumi ingest on miss
                 ├── search_bangumi → catalog points for resolved bangumi_id
                 ├── search_nearby  → catalog geo retrieval (PostGIS on Neon)
                 ├── plan_route     → catalog route ordering
                 └── web_search / translate → attributed research / title translation
              → answer + tool call records

A single agent handles planning and tool dispatch. Selected-point routes bypass the agent entirely.

resolve_anime is self-evolving: on first query for an unknown title it fetches metadata from Bangumi.tv, upserts it into the database, and all future queries hit the local DB.

Features

• Conversational search — ask in Japanese, English, or Chinese; the agent handles intent
• Self-evolving anime catalog — DB-first with Bangumi.tv API write-through on miss
• Geo retrieval — find pilgrimage spots near any coordinate or station name
• Route planning — nearest-neighbor ordering with optional user-selected points
• Generative UI — three-column layout with chat panel + interactive result panel
• Edge auth — Neon Auth (Better Auth) JWT (magic-link) enforced at Cloudflare Worker against the branch JWKS
• Eval harness — 50+ plan-quality cases across 3 locales (packages/eval)

Quick Start

# Install dependencies
pnpm install

# Run the web app locally
make dev-local

# Run every package's lint, typecheck and tests
make check-full

Database Migrations

Neon schema changes are declared in packages/pi-session-neon/src/contract.prisma and versioned as one Prisma 8 chain under packages/pi-session-neon/migrations/. Its emitted artifacts must be regenerated in the same change. No Worker keeps a schema of its own: the contract that chain generates is the only map of this data plane (#1633). supabase/ is an archived historical Supabase migration tree (issue #1000); it is not applied and is not a source for new Neon tables.

make db-new NAME=x     # scaffold a migration in the chain
make db-lint           # artifact integrity and a connected graph
make db-status         # the migration path and what is pending

There is no local apply target: the migrator Worker holds the only database credential, and CD is the only thing that applies a migration.

See docs/ops/migrations.md for the boundary, CI gates, and deploy order. Apply migrations in a dedicated deploy step, not at application startup.

Environment

Required (edge Worker / local serve):

Variable	Purpose
AGENT_SVC_DATABASE_URL	Neon agent_svc role DSN — the required agent data-plane connection (#912), bound on the edge Worker and read by the native agent tier
MIMO_API_KEY	Primary model provider key

Worker edge: NEON_AUTH_JWKS_URL (the edge's ONLY identity source — AUTH-2 #950; verifies Neon Auth EdDSA JWTs against the branch JWKS; production stays unset/fails closed until its branch is provisioned). Catalog/users/jobs also need their Neon DSNs — see docs/ops/deployment.md.

Web (apps/web): VITE_NEON_AUTH_BASE_URL — the Better Auth client origin (login UI + JWT exchange); VITE_TURNSTILE_SITE_KEY, VITE_SHOWCASE_MODE — see apps/web/.env.example.

Optional: SERVICE_HOST, SERVICE_PORT, OBSERVABILITY_*, DEFAULT_AGENT_MODEL

See .env.example for defaults.

Example Usage

HTTP (authenticated):

curl -N -X POST https://seichijunrei.zhenjia.org/v1/chat \
  -H 'Authorization: Bearer <neon_auth_jwt>' \
  -H 'Content-Type: application/json' \
  -H 'x-locale: ja' \
  -d '{"messages":[{"role":"user","parts":[{"type":"text","text":"吹響の聖地"}]}]}'

Repository Map

• workers/catalog/ — Cloudflare Worker: anime catalog API + data platform (TypeScript)
• workers/users/ — Cloudflare Worker: user-domain data service (/v1/users/*)
• packages/contract/ — shared oRPC/zod contract (catalog ↔ agent ↔ users)
• apps/web/ — TanStack Start SSR web app (the only browser surface)
• workers/edge/ — Cloudflare Worker entrypoint for auth and /v1 routing
• supabase/ — legacy compatibility migrations and Supabase project assets (auth retired to Neon, AUTH-2 #950)
• docs/ — architecture, ops runbooks, iteration artifacts, and implementation plans
• Makefile, package.json — root tooling entrypoints; workers/edge/wrangler.toml (edge Worker config) lives beside its code

Docs

• Architecture — full system design reference
• Deployment — Cloudflare Workers deploy guide
• Migrations — the Prisma chain's authority and what may not apply it
• Ops docs — operational runbooks and environment procedures
• Iteration artifacts — task plans, progress logs, and findings by iteration
• Implementation plans (archive) — historical execution plans (flat plans/ no longer accepts new files)
• Design specs — active product/architecture specifications
• Agent guide — monorepo layout, commands, and cross-stack guardrails