Is it secure to expose logs to an agent?

Yes, when following the skill's guidelines. It emphasizes using development-only environments, redacting sensitive PII, and ensuring logs are only accessible within the local dev context.

What is the best way to expose logs to an agent?

It depends on the project: use development-only endpoints for web apps, log files for CLI tools, or unified query scripts for monorepos.

How does the agent know how to find these logs?

The skill includes a phase for updating CLAUDE.md or AGENTS.md with specific schema details and example commands that the agent can execute to retrieve data.

Why do AI agents need specialized telemetry?

Standard logs are often unstructured or locked in cloud consoles. This skill makes logs queryable directly by the agent, allowing it to see exactly what happened during execution without human intervention.

Agent Telemetry

Name: Agent Telemetry
Author: petekp

bypetekp

0•

분석 및 모니터링

Empowers AI coding agents to debug runtime behavior through structured logging and queryable telemetry endpoints.

Agent Telemetry is a specialized skill designed to bridge the gap between static code analysis and runtime execution for AI agents. By implementing structured JSON logging, exposing accessible telemetry endpoints, and documenting specific query commands in CLAUDE.md or AGENTS.md, this skill ensures that coding agents can independently diagnose issues, trace requests, and verify state transitions. It provides a systematic workflow for auditing existing logs, adding correlation IDs, and setting up agent-friendly query tools, ultimately reducing the time spent on manual debugging and increasing the autonomy of AI-driven development.

주요 기능

01Implement structured JSON logging with request correlation IDs

02Expose logs via dev-only endpoints, local files, or CLI query tools

03Audit existing logging infrastructure and identify telemetry gaps

04Validate end-to-end observability workflows for AI agent usability

050 GitHub stars

06Generate agent-optimized documentation with copy-pasteable commands

사용 사례

01Debugging elusive runtime errors that aren't visible through static code analysis

02Enabling agents to trace requests across service boundaries using correlation IDs

03Setting up observability for new applications to ensure they are 'agent-ready' for future maintenance

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add petekp/agent-skills agent-telemetry

For use in Claude.ai and ChatGPT

Download Skill