Use Longwake with an AI agent

Everything you can see here is also available as plain data.

No key is needed and the API is read-only. A good starting point for any model is /llms.txt.

API

GET /api/people: filter with topic, open_to, contactable=1, org and q. Responses are JSON, or Markdown with format=md.
GET /api/people/{slug}: one full profile, including how to reach the person.
GET /api/orgs and GET /api/topics.

Example: people working on AI-enabled coups who want collaborators, as Markdown:

https://longwake.org/api/people?topic=ai-enabled-coups&open_to=collaborating&format=md

MCP server

The repository includes a small MCP server with four read-only tools: search_people, get_person, search_orgs and list_topics. With Claude Code:

claude mcp add longwake -e DIRECTORY_URL=https://longwake.org -- node /path/to/this/repo/mcp/server.mjs

Or in a desktop client's MCP config:

{
  "mcpServers": {
    "longwake": {
      "command": "node",
      "args": [
        "/path/to/this/repo/mcp/server.mjs"
      ],
      "env": {
        "DIRECTORY_URL": "https://longwake.org"
      }
    }
  }
}

Etiquette

Agents draft notes; people send them. Nothing here sends messages, and agents shouldn't either.
Respect each person's contact setting. If they ask for a note, make it short and specific.