Skip to content

CHANGELOG.md

Latest commit

 

History

History
96 lines (74 loc) · 2.95 KB

CHANGELOG.md

File metadata and controls

96 lines (74 loc) · 2.95 KB

Changelog

All notable changes to the LangSmith Fetch Skill will be documented in this file.

The format is based on Keep a Changelog, and this project adheres to Semantic Versioning.

[0.1.0] - 2025-12-24

🎉 Initial Release

The first AI observability & debugging skill for Claude Code!

Added

Core Features

  • ✅ Automatic debugging - Claude autonomously fetches and analyzes LangSmith traces
  • ✅ Four core workflows:
    1. Quick Debug Recent Activity (5-minute traces)
    2. Deep Dive Specific Trace (by trace ID)
    3. Export Debug Session (save to organized folders)
    4. Error Detection (find and categorize failures)

Skill Capabilities

  • ✅ Fetch recent traces with time-based filtering
  • ✅ Analyze specific trace by ID
  • ✅ Export sessions to files with metadata
  • ✅ Detect and categorize errors
  • ✅ Review tool calls and results
  • ✅ Check memory operations (LTM)
  • ✅ Track token usage and costs
  • ✅ Compare agent performance
  • ✅ Identify bottlenecks
  • ✅ Suggest optimizations

Documentation

  • ✅ Complete SKILL.md with YAML frontmatter and detailed workflows
  • ✅ Professional README.md with installation and usage guides
  • ✅ CONTRIBUTING.md with contribution guidelines and templates
  • ✅ MIT LICENSE
  • ✅ This CHANGELOG.md

Examples & Guides

  • ✅ Response format examples for each workflow
  • ✅ Common use cases (Agent Not Responding, Wrong Tool Called, Memory Not Working, Performance Issues)
  • ✅ Troubleshooting guide
  • ✅ Best practices for debugging
  • ✅ Quick reference command guide

Integration

  • ✅ Seamless Claude Code integration
  • ✅ Model-invoked activation (Claude decides when to use)
  • ✅ PowerShell and Bash command examples
  • ✅ Environment variable setup guide

Requirements

  • langsmith-fetch CLI (>= 0.1.0)
  • LANGSMITH_API_KEY environment variable
  • LANGSMITH_PROJECT environment variable

Activation Keywords

Claude automatically activates this skill when users mention:

  • "Debug my agent"
  • "What went wrong?"
  • "Show me recent traces"
  • "Check for errors"
  • "Analyze memory operations"
  • "Review agent performance"
  • "What tools were called?"

Authors

  • Ahmad Othman Ammar Adi - Initial work - @OthmanAdi

Acknowledgments

  • LangChain team for the excellent langsmith-fetch CLI
  • Anthropic for Claude Code and the Skills framework
  • The AI observability community

[Unreleased]

Planned Features

  • Enhanced multi-agent orchestration debugging
  • Cost tracking and optimization suggestions
  • Performance profiling workflows
  • Custom export formats
  • Team collaboration features

Note: This is the first AI observability skill for Claude Code. We're excited to see how the community uses and improves it!

For full details on each release, see the Releases page.