ACI.dev

The platform provides a unified way for AI agents to access more than 600 external services through either direct function calls or a Model‑Context‑Protocol (MCP) server. It includes built‑in multi‑tenant authentication, OAuth flows, and secret management, allowing developers to grant agents intent‑aware access while enforcing granular, natural‑language permission boundaries. Dynamic tool discovery and logging let agents select appropriate integrations without overloading LLM context windows.

It is aimed at developers building agentic IDEs, custom AI assistants, or automated DevOps workflows that need reliable, secure tool usage across many services. The system works with any LLM framework or agent architecture and can be deployed on platforms such as Vercel, Supabase, or Cloudflare. A lightweight Python SDK offers direct function‑calling, while the unified MCP server provides a consistent API surface for larger deployments.

All components are open source under the Apache 2.0 license, including the backend, developer portal, and the extensive catalog of pre‑built integrations. This enables teams to avoid vendor lock‑in, reuse the same authentication and permission logic across projects, and extend the platform with additional tools as needed.