Agent Trace Auditor

lazymac2x/agent-trace-auditor-api

A remote MCP server running on Cloudflare Workers that audits agent trace data, accessible via streamable HTTP at api.lazy-mac.com/agent-trace-auditor. The implementation combines a REST API with MCP's JSON-RPC 2.0 protocol, letting you hook into trace auditing without running local infrastructure. Reach for this when you need to analyze or validate agent execution traces from Claude and want the auditing logic handled server-side. The Cloudflare Workers deployment means low latency and global distribution, though you'll want to check the repository for specifics on what trace formats it expects and what audit operations it exposes.

tools: 3
transport: HTTP
last updated: May 16, 2026

View on GitHub

Install

From registry metadata

$ npx add-mcp https://api.lazy-mac.com/agent-trace-auditor/mcp --name agent-trace-auditor

Run in your terminal. add-mcp asks which agents to add it to (Claude Code, Codex, Cursor and more). Add -g to install it for every project.

Review the command, arguments and environment values before installing. MCP servers run with your local permissions.

Other ways

Claude Code

$ claude mcp add --transport http agent-trace-auditor https://api.lazy-mac.com/agent-trace-auditor/mcp

Run in your terminal. Add --scope user to make it available in every project.

Tools

Verified live · 3 tools

Verified live against the running server on Nov 6, 2026.

audit_traceFull audit of an agent execution trace. Detects infinite loops, duplicate calls, tool errors, cost overruns, schema violations, and latency bottlenecks. Returns verdict (pass/review/block), score 0-100, and detailed findings.3 params

Full audit of an agent execution trace. Detects infinite loops, duplicate calls, tool errors, cost overruns, schema violations, and latency bottlenecks. Returns verdict (pass/review/block), score 0-100, and detailed findings.

Parameters* required

budget_usdnumber

Cost cap in USD (default 5.0)

tool_schemasobject

Per-tool required field schemas for input validation

trace*array

Agent execution steps. Each step: {tool, input, output, model, ts, latency_ms, error}

quick_verdictFast pass/review/block verdict with score and issue counts. Use in CI gates.2 params

Fast pass/review/block verdict with score and issue counts. Use in CI gates.

Parameters* required

budget_usdnumber

trace*array

Agent execution steps

diff_tracesCompare two agent traces to detect regressions. Returns score delta, cost delta, added/removed tools, new/resolved issue types.3 params

Compare two agent traces to detect regressions. Returns score delta, cost delta, added/removed tools, new/resolved issue types.

Parameters* required

budget_usdnumber

trace_a*array

Baseline (previous) trace

trace_b*array

New trace to compare against baseline

Readme