MyApi

MyApi addresses a specific security gap in AI agent setups: raw API credentials scattered across local .env files, no audit trail of what an agent actually did, and no way to revoke one agent’s access without rotating every key you own. It acts as a gateway between your sensitive service credentials and the agents that use them — connect a service once, then issue scoped, revocable tokens per agent instead of handing out raw keys.

Designed specifically for the growing ecosystem of AI agent tools (OpenClaw, Hermes, Claude Code, and others), MyApi centralizes credential management with per-agent scoping and a complete audit trail of every action taken through it, so access can be reviewed and revoked precisely rather than all-or-nothing.

AGPL-3.0 licensed and self-hostable via Docker, MyApi also offers a fully managed, zero-ops hosted instance at myapiai.com for users who want to skip server setup entirely while running the same underlying platform.

What You Get

• Centralized credential management — connect a service once instead of configuring every agent individually
• Scoped, revocable tokens issued per agent, so access can be cut off without rotating every key
• A complete audit trail of every action taken by an agent through the gateway
• Self-hosted deployment via Docker, or a fully managed zero-ops hosted instance

Common Use Cases

• Giving multiple AI agents (OpenClaw, Hermes, Claude Code) scoped access to the same underlying services without sharing raw API keys
• Revoking one compromised or unneeded agent’s access without rotating credentials shared by every other agent
• Auditing exactly what actions an AI agent took using a connected service’s credentials
• Centralizing credential storage instead of scattering API keys across multiple agents’ local .env files

Under The Hood

Architecture MyApi sits as a gateway layer between connected services and the AI agents that need to use them, issuing per-agent scoped tokens rather than exposing raw service credentials directly to agent processes. This indirection is what enables both fine-grained revocation (kill one agent’s token without affecting others) and a complete audit trail (every request routes through the gateway, so every action is logged centrally rather than scattered across each agent’s own logging).

Tech Stack Node.js (18+) for the backend, with a React-based dashboard app (src/public/dashboard-app) for managing connections and tokens, Docker support for self-hosted deployment, and CI configured via GitHub Actions.

Code Quality The project runs CI on every change and ships a dedicated dashboard UI rather than a config-file-only tool, suggesting attention to usability alongside the core security model; as a young, low-star project, broader community validation is still limited.

What Makes It Unique Most AI agent setups treat credential management as each agent’s own responsibility (raw keys in environment variables); MyApi centralizes it as a gateway with per-agent scoping and audit logging, addressing the specific failure mode of “security only as strong as your last .env file” that becomes acute once multiple agents need access to the same sensitive services.

Self-Hosting

Licensing Model AGPL-3.0 licensed — fully open source with no license key for self-hosting.

Self-Hosting Restrictions None found; the same code is available to self-host via Docker.

Cloud vs Self-Hosted A fully managed, always-updated hosted instance is available at myapiai.com with a free sign-up option, for users who want to skip server setup; self-hosting gives full control over where credentials and audit logs are stored.

License Key Required No.