Executor
Executor connects all the services an AI agent uses into one place, and lets agents build and run code-based tools
The Problem
People using AI coding agents and assistants need those agents to reach services like Gmail, GitHub, Linear, and Stripe, but connecting each account per agent or embedding credentials in chat or code is risky and repetitive. When they want something more than a single tool call, such as a scheduled job, a webhook-triggered action, or a small dashboard, there is no obvious place for that logic to live or keep running. Existing options like flow builders (n8n, Zapier) require building automations visually rather than letting the agent write code, while other connector platforms mainly handle authentication without letting users bring their own integrations or build full apps on top.
The Solution
Executor is described as an 'agents cloud': you connect accounts like Gmail, Calendar, GitHub, Linear, Slack, Notion, Stripe, Drive, PostHog, Sentry and X once through a secure link, and every agent you use (Claude Code, Codex, Cursor, ChatGPT, Gemini) can then reach them through a single MCP server. Agents can call existing tools or write new ones as plain JavaScript/TypeScript, and apps built this way can combine tools, durable multi-step workflows, scheduled cron jobs, webhooks, persistent storage, and a React-based UI page with its own URL. The page walks through a concrete example (a 'Daily brief' app) showing an entry point file, tool definitions, an OAuth provider declaration, a SQL migration, a background workflow, a SKILL.md instruction file, and a React UI, all deployable to Executor Cloud or run via desktop app, CLI, or self-hosted server.