For AI agents
Install one skill and any coding agent, Claude Code, Cursor, Codex, Windsurf, can publish, schedule and measure across every network. Prefer MCP or plain REST? Those work too.
npx skills add postlake/postlake-mcp --all
Adds five skills. Accounts, publish, schedule, media, analytics. Set a POSTLAKE_API_KEY and you're live.
One command installs five skills into your coding agent. No glue code.
Point any MCP client at our server and approve once over OAuth.
Call it from any language or framework. LangChain, CrewAI, OpenAI tools.
Step-by-step for the popular clients:
Let Claude post to all your socials through PostLake.
Let ChatGPT post to all your socials through PostLake.
Let Cursor post to all your socials through PostLake.
Let Gemini post to all your socials through PostLake.
Let GitHub Copilot post to all your socials through PostLake.
Connect PostLake's hosted MCP server to your AI assistant (Claude, ChatGPT, Cursor, Gemini, Copilot and others), approve it once with OAuth, then just ask it to post. No API key to paste and no code required.
Any MCP-capable assistant. Claude, ChatGPT, Cursor, Gemini CLI, VS Code / GitHub Copilot, and any other MCP host, plus any agent framework like LangGraph or CrewAI via the REST API.
No. The MCP server is no-code: connect it to your AI host and ask in plain language. If you're building your own agent runtime, the REST API and idempotency keys are there too.
Only approve clients you trust. OAuth avoids pasting an existing API key into the client, idempotency keys prevent duplicate posts, publishing limits are enforced, and you can revoke access instantly from Agents.
There's a free tier, 20 credits a month, no card, then usage-based credits at one per post (X costs more). The MCP server and agent access are included on every plan.