InsForge is a backend platform that coding agents operate directly. It bundles authentication, a Postgres database, S3-compatible storage, edge functions, a model gateway, container compute, and site deployment, so an agent can build and ship a full-stack app without a person wiring those services together.
Agents reach InsForge through one of two interfaces. An MCP server (self-hosted and cloud) exposes its operations as tools any MCP-compatible agent can call. A CLI paired with Skills (cloud only) lets agents invoke the same operations from the terminal. Through either interface an agent reads backend context (documentation, schemas, deployed functions, bucket contents, auth configuration, and runtime logs) and configures primitives: it deploys edge functions, runs database migrations, creates storage buckets, and sets up auth providers.
InsForge runs as a hosted cloud service or self-hosted with Docker Compose, with one-click templates for Railway, Zeabur, Sealos, RepoCloud, and ZopDay. Self-hosted storage defaults to the local filesystem and can be backed by bundled MinIO or RustFS, or by an external S3-compatible store such as AWS S3, Wasabi, or R2, which also enables an S3-compatible gateway for standard S3 clients.
Features
- Authentication: user management, sign-up and login with OAuth providers, and sessions
- Database: a Postgres relational database that agents migrate and inspect
- Storage: S3-compatible file storage, with local, bundled, or external S3 backends when self-hosted
- Model gateway: an OpenAI-compatible API in front of multiple LLM providers
- Edge functions: serverless code that agents deploy and debug from runtime logs
- Compute: long-running container services (private preview)
- Site deployment: builds and hosts a site at a live URL
- MCP server: exposes backend operations as tools for any MCP-compatible coding agent
- CLI and Skills: a command-line interface with Skills for agents working in the terminal (cloud only)
- Backend context for agents: documentation, schemas, metadata, and logs an agent reads to write, verify, and debug code
