jurupari

Personal knowledge base or second brain with semantic search. PostgreSQL + pgvector + MCP + HTTP API.

Why? We are done with markdown, they are nice, until you have millions of them.

The name

Jurupari is an indigenous Amazonian legislator and spirit who established social laws, rituals, and sacred traditions for tribes.

Workflwo

The workflow is simple, you deploy it in the place is better for you, like AWS, Railway... and plug the MCP to your Claude, GPT, Claude Code, OpenClaw, Hermes...

Local dev

cp .env.example .env          # fill DATABASE_URL, OPENAI_API_KEY, JURUPARI_TOKEN
docker compose up -d          # starts postgres with pgvector
pnpm install
pnpm db:push                  # push schema (dev) — use db:migrate in prod
pnpm dev:api                  # HTTP API on :3000
pnpm --filter @jurupari/mcp build && node packages/mcp/dist/index.js  # MCP server

MCP config

Remote SSE — Claude.ai web, any client without local access

{
  "mcpServers": {
    "jurupari": {
      "type": "sse",
      "url": "https://your-url/mcp/sse?token=your-secret-token"
    }
  }
}

No token → level-0 read-only. With valid token → full access (level-1).

Local stdio — Claude Desktop, Claude Code, Cursor (lower latency, always full access)

{
  "mcpServers": {
    "jurupari": {
      "command": "node",
      "args": ["/absolute/path/to/jurupari/packages/mcp/dist/stdio.js"],
      "env": {
        "DATABASE_URL": "postgresql://...",
        "OPENAI_API_KEY": "sk-...",
        "JURUPARI_TOKEN": "your-secret-token"
      }
    }
  }
}

Pola can use either: HTTP REST directly (/memories), or the SSE MCP endpoint.

Access levels

Level	Read requires	Write requires	Use case
0	nothing extra	token	general notes, non-sensitive
1	token	token	personal data, process numbers
2	(future)	(future)	reserved for strictest data

HTTP API

GET    /health
POST   /memories/search   body: { query, category?, limit? }
GET    /memories?category=legal
GET    /memories/:id
POST   /memories          Authorization: Bearer <token>
PATCH  /memories/:id      Authorization: Bearer <token>
DELETE /memories/:id      Authorization: Bearer <token>

Deploy to Railway as an example

1. Push repo to GitHub
2. New Railway project → connect repo
3. Add env vars: DATABASE_URL (Railway Postgres addon), OPENAI_API_KEY, JURUPARI_TOKEN
4. Railway reads railway.toml automatically