Vigil Connect

AI is brilliant but forgetful. Give it your firm’s decision record.

Vigil is the system of record for organizational reasoning. Connect makes its decisions, assumptions, and cited evidence available to Claude Desktop, Cursor, and other MCP clients. Your AI can retrieve what the team believed, test new information against it, and cite the exact source material behind the answer. Vigil keeps monitoring the record after the conversation ends.

What your AI can ask Vigil

Read the decision record

What does this workspace currently believe about a company, deal, or thesis? The answer includes the decision, its current assumptions, recent changes, and the cited evidence behind anything at risk.

Check for conflicts

Does a new statement conflict with what the team has already established? Vigil returns a verdict tied to the specific decisions, assumptions, source claims, and verbatim evidence involved.

Verify a claim

Supported, contradicted, or insufficient evidence — judged only against the material in your workspace, with citations back to the exact observed source version.

Search the evidence

Find collected claims with verbatim quotes, source attribution, observation dates, version references, and staleness signals.

Every response is grounded in your workspace and carries citations to your own source material. Vigil reports the evidence match; it does not invent a certainty score or turn the record into investment advice.

Setup

  1. Open Vigil → Connect and create an API key (included with every Vigil engagement; testers get complimentary access).
  2. Add the snippet for your client, with your key in place of YOUR_VIGIL_KEY.
  3. Restart the client — the four Vigil tools appear automatically.

Claude Desktop — claude_desktop_config.json

{
  "mcpServers": {
    "vigil": {
      "command": "npx",
      "args": [
        "-y", "mcp-remote", "https://trustvigil.com/api/connect/mcp",
        "--header", "Authorization: Bearer YOUR_VIGIL_KEY"
      ]
    }
  }
}

Cursor — .cursor/mcp.json

{
  "mcpServers": {
    "vigil": {
      "url": "https://trustvigil.com/api/connect/mcp",
      "headers": { "Authorization": "Bearer YOUR_VIGIL_KEY" }
    }
  }
}

What Connect shares, plainly

A Connect key gives the AI client you configure read access to one workspace's decisions, assumptions, and collected evidence. The tool registry and permissions are closed: a key can call only the four documented read tools. When a tool is called, returned content enters the third-party model's context — governed by that provider's terms, not Vigil's. Calls are rate-limited and recorded in a content-free call trail. Keys can be revoked at any time in the app, effective immediately.

Connect is included with every Vigil engagement — there is no separate API product. Access is currently limited to customer workspaces and invited testers.