Skip to content
csl
All skills

InsForge MCP

MCP servers

MCP server giving coding agents an open-source backend: Postgres, auth, storage, edge functions, model gateway, deploys.

by InsForgeInsForge/insforge
Stars13.2k
Forks1.2k
Watchers36
Open issues227
LicenseApache-2.0
Last pushtoday

Install

install

$ npx @insforge/install --client claude-code --env API_KEY=YOUR_API_KEY --env API_BASE_URL=http://localhost:7130

Update

update

$ npx @insforge/install --client claude-code --env API_KEY=YOUR_API_KEY --env API_BASE_URL=http://localhost:7130

Installer wires up `@insforge/mcp@latest`, so re-running install (or restarting the client) fetches the newest server. Add `--dev` to pin the @insforge/mcp@dev channel.

When to use it

  • You want a coding agent to provision and operate a real backend (DB, auth, storage, functions) end-to-end, not just write app code.
  • You are self-hosting via Docker Compose or using the insforge.dev cloud and need to connect it to Claude Code over MCP.
  • You want the agent to read live backend state — schemas, deployed functions, bucket contents, runtime logs — to verify and debug what it built.
  • You are migrating off Supabase or building a full-stack app and want one MCP-driven backend platform.

Commands & flags

fetch-docs
Pull InsForge instructions, schemas, and API references so the agent knows how to operate the backend.
npx @insforge/install --client claude-code
Wire the @insforge/mcp server into Claude Code with your API_KEY and API_BASE_URL.
docker compose -f docker-compose.prod.yml up
Self-host an InsForge instance locally; web UI and MCP connect flow at http://localhost:7130.
Database / migrations
Run Postgres migrations and read schemas plus metadata directly from the agent.
Edge Functions / Storage
Deploy serverless edge functions and create S3-compatible storage buckets.
Model Gateway
Call an OpenAI-compatible API across multiple LLM providers through the backend.

How to use it

This is an MCP server, not a skill/plugin. It needs a running InsForge backend (self-hosted or cloud) plus an API key.

  1. Stand up a backend — self-host:
git clone https://github.com/InsForge/InsForge.git
cd insforge
cp .env.example .env
docker compose -f docker-compose.prod.yml up

Open http://localhost:7130, follow the connect flow, and grab your API_KEY. (Or skip Docker and use the cloud at insforge.dev.)

  1. Register the MCP server with Claude Code (command from the InsForge/insforge-install repo):
npx @insforge/install --client claude-code \
  --env API_KEY=YOUR_API_KEY \
  --env API_BASE_URL=http://localhost:7130

Restart the client to load the server. Add --dev to use the @insforge/mcp@dev channel.

  1. Verify by prompting the agent to call the fetch-docs tool:
I'm using InsForge as my backend platform, call InsForge MCP's fetch-docs tool to learn about InsForge instructions.

Why it earns its place

Open-source (Apache-2.0) all-in-one backend purpose-built for agentic coding — one MCP surface covering Postgres, auth, S3-compatible storage, edge functions, an OpenAI-compatible model gateway, compute, and site deployment, so an agent can operate the backend like a backend engineer instead of you hand-wiring services. Backed by the InsForge org with published @insforge/sdk, @insforge/mcp, and @insforge/install packages on npm, Vercel OSS membership, and a Trendshift listing, and it ships self-hosted (Docker Compose) and cloud paths plus migration tooling from Supabase.

Try it

claude

> Using the InsForge MCP, create a posts table with RLS, deploy an edge function to handle uploads to a new storage bucket, then show me the schema.