Agentic Coop DB banner
fheinfling fheinfling

Agentic Coop DB

Data community

Description

Auth gateway for a shared PostgreSQL database for agentic use—parameterized SQL forwarding with API keys, RLS, and pgvector

Installation

This entry records only its repository, not the path inside it, so there is no exact command to give. Open the source below and copy the folder into ~/.claude/skills/, or the file into ~/.claude/agents/.

README

Agentic Coop DB — auth gateway for shared PostgreSQL

[![CI](https://github.com/fheinfling/agentic-coop-db/actions/workflows/ci.yml/badge.svg)](https://github.com/fheinfling/agentic-coop-db/actions/workflows/ci.yml) [![License](https://img.shields.io/badge/License-Apache_2.0-blue.svg)](LICENSE)

**Ever felt the need to connect your AI agents to an actual database?** Agentic Coop DB does exactly that — give every agent a secure SQL interface over HTTPS, backed by real Postgres, with nothing more than an API key.

Multiple AI agents working on the same project need a shared place to store and query structured data — but you can't expose Postgres on the public internet. Agentic Coop DB lets your agents collaborate on a remote Postgres + pgvector instance using nothing but an HTTPS URL and an API key. Share the results your tokens generated, build on each other's work, and query it all with plain SQL — no new query language, no ORM lock-in.

**Status:** v0.3 — single-node, container-first, ARM64-friendly. **License:** Apache-2.0.


What it is

A thin auth gateway in front of PostgreSQL 16 + pgvector that does four jobs:

  1. Authenticate the caller via a workspace-scoped API key.
  2. Authorize by attaching a Postgres role to the request transaction (SET LOCAL ROLE) — Postgres itself decides what the key can run.
  3. Forward the SQL with parameterized binding and a statement timeout.
  4. Audit every call.

If you can write SQL, you can use it. `SELECT`, `INSERT`, `UPDATE`, `DELETE`, `CREATE TABLE`, `CREATE USER`, `GRANT`, pgvector ops — all forwarded.

What it is not

  • Not a new query language, ORM, or schema migrator.
  • Not realtime / websocket subscriptions.
  • Not object storage.
  • Not a serverless function runtime.
  • Not multi-region or HA — single-node only in v1.
  • Not a web UI — CLI + curl + your own app.
  • Not SSO — API keys only.

Getting started

Option A — Local development

git clone https://github.com/fheinfl