Audit Every AI Tool Call with the OE MCP Action Log

Every time Claude, Cursor, or Windsurf calls a connector through OE MCP — a SQL query, a Slack post, a GitHub file read — it is automatically recorded. No setup. No config changes. Just open oe-mcp-log.json or ask Claude to show you the log, and see exactly what happened, when, and on which connector.

What You Need

• OE MCP installed — run npx -y @openenthrium/oe-mcp or follow the setup guide
• Claude Code, Cursor, Windsurf, or any MCP-compatible AI assistant
• At least one connector configured in your oe-mcp.json

Why This Matters

When your AI assistant queries your production database, writes to a file, or posts a message to Slack, you want a record. Not because you don't trust the AI — but because debugging is faster, compliance is cleaner, and "what did it actually do?" is a question you will ask.

OE MCP answers it automatically.

What Gets Logged

Every connector tool call is appended to oe-mcp-log.json in the same folder as your oe-mcp.json. Each entry records:

• ts — ISO timestamp of the call
• connector — which connector was used (e.g. my-postgres)
• tool — which tool was called (e.g. query, post_message)
• input — the exact parameters passed
• result — ok or the error message if it failed

Example — log entries from a real session

[
  {
    "ts": "2026-08-12T10:15:22.401Z",
    "connector": "my-postgres",
    "tool": "query",
    "input": { "sql": "SELECT * FROM orders WHERE status = 'pending' LIMIT 20" },
    "result": "ok"
  },
  {
    "ts": "2026-08-12T10:16:04.882Z",
    "connector": "my-slack",
    "tool": "post_message",
    "input": { "channel": "#alerts", "text": "20 pending orders found. Review needed." },
    "result": "ok"
  },
  {
    "ts": "2026-08-12T10:17:31.114Z",
    "connector": "my-github",
    "tool": "read_file",
    "input": { "path": "src/orders/processor.ts" },
    "result": "ok"
  }
]

The Two Log Tools

OE MCP exposes two built-in tools for working with the log. They are available in every session — no extra configuration needed.

Ask Claude in plain language

# You say:
"Show me the action log"
# Claude calls log_list → returns all recent connector activity

"Show me the last 5 actions"
# Claude calls log_list with limit: 5

"Clear the action log"
# Claude calls log_clear → log is wiped

What You Can Do With It

Where the Log File Lives

OE MCP creates oe-mcp-log.json automatically next to your config file — no setup required:

/your/config/folder/
├── oe-mcp.json         # your connector config
├── oe-mcp-memory.json  # persistent memory
└── oe-mcp-log.json     # action log — auto-created on first call

It is a plain JSON array you can read, parse, or back up at any time.

Setting Up OE MCP

The Action Log is built into every OE MCP session — no extra configuration needed. If you haven't connected OE MCP to your AI assistant yet, it takes under two minutes.

Step 1 — Create your oe-mcp.json config file:

{
  "connectors": [
    {
      "name": "my-postgres",
      "type": "postgresql",
      "host": "localhost",
      "port": 5432,
      "database": "mydb",
      "user": "postgres",
      "password": "secret"
    },
    {
      "name": "my-slack",
      "type": "slack",
      "botToken": "xoxb-xxxxxxxxxxxx"
    }
  ]
}

Step 2 — Add to your AI app's MCP config (Claude Code, Cursor, Windsurf, Codex, Claude Desktop) — macOS / Linux:

{
  "mcpServers": {
    "oe-mcp": {
      "type": "stdio",
      "command": "npx",
      "args": ["-y", "@openenthrium/oe-mcp", "--stdio", "/path/to/oe-mcp.json"]
    }
  }
}

Windows — use npx.cmd instead of npx:

{
  "mcpServers": {
    "oe-mcp": {
      "type": "stdio",
      "command": "npx.cmd",
      "args": ["-y", "@openenthrium/oe-mcp", "--stdio", "C:\\path\\to\\oe-mcp.json"]
    }
  }
}

Reload your AI app — every connector call is logged automatically from the first request.