skill-workshop

To turn Claude Code session patterns into reusable skills, analyze saved JSONL session logs to discover recurring explanations and tool sequences, then generate draft SKILL.md files in the .claude/skills directory. This process mines tribal knowledge from project history.

Session history mining for workflow discovery is the process of scanning Claude Code JSONL logs to uncover repeated explanations, recurring tool chains, and workaround patterns. It identifies tribal knowledge that should become structured, ranked skill candidates.

Extract tribal knowledge from JSONL session logs by using a session-analyzer agent to evaluate project session history. The analysis identifies repeated workaround patterns and ranks them as skill candidates with evidence, frequency, and suggested skill types.

Skill generation from session history requires a valid project path with saved JSONL session logs in the session directory. The system validates the session directory before performing delegated extraction to ensure session analysis can occur.

The best way to organize draft skills discovered from session analysis is to rank candidates by frequency and evidence, categorize them as knowledge, workflow, or gotcha skill types, and generate controlled SKILL.md drafts in .claude/skills.

You should not use session analysis for skill generation if your project lacks saved JSONL session logs or if your team has not accumulated repeated explanations and workaround patterns worth converting into durable internal skills.