Audit MCP Tool Call Logs

MCP Call Auditor is a free, MIT-licensed agent skill that reviews your AIML.chat MCP tool-call logs for safety — confirming every write was approved, flagging over-permissive tools, and surfacing unusual call patterns. It gives you the audit trail that makes agent actions trustworthy.

What it does

The skill reads the MCP call log and checks the things that matter for safety: that every mutating call went through write-confirm and was approved by a human, that no tool is reaching beyond its intended scope, and that call volume and targets look normal rather than anomalous. It groups calls by tool and agent, flags any write that looks risky, and summarizes which tools are used most so you can prune what you don't need. For teams connecting CRMs, billing, or repos to agents, this is the periodic safety review that keeps least-privilege honest. The output is an audit summary with anything that needs attention.

The skill

name: mcp-call-auditor
Use to review MCP tool-call logs for safety and scope.

Steps:
1. Pull the MCP call log from AIML.chat.
2. Confirm every mutating call was approved via write-confirm.
3. Flag over-permissive tools and anomalous call patterns.
4. Output an audit summary with items needing attention.

How to use it

Copy the SKILL.md into your agent framework's skills folder and run it on exported logs, or review the call log in AIML.chat's MCP section. It pairs with the MCP auth setup skill to tighten scopes and the MCP server tester to verify servers. The MCP docs explain how write-confirm and the call log work.

Ready to audit? Start free, connect a server, and keep its actions accountable — see pricing for limits.