story-miner

To extract publishable development stories from Claude Code session history, this Skill mines preprocessed session data to assemble narratives with grounded quotes and provenance pointers. It scores candidates for narrative value and deduplicates similar incidents before generating the final story artifacts.

Story mining is the process of analyzing Claude Code session logs to produce publishable development narratives. It operates on preprocessed data files to extract grounded quotes, deduplicate incidents, and generate documentation suitable for onboarding and learning.

You can convert Claude Code session logs into markdown documentation by processing preprocessed data through the story mining pipeline. It outputs human-readable markdown files, specifically stories.md and digest.md, containing assembled narratives with provenance tracking.

Yes, the story mining process requires specific preprocessed data formats to function correctly. It operates on preprocessed.json, candidates.jsonl, and stories.jsonl files to assemble narratives, score candidates, and deduplicate clusters before writing final output artifacts.

You can redact sensitive information while preserving traceable quotes in development stories using built-in redaction safeguards. The Skill ensures safe redaction of internal logs while maintaining provenance pointers that allow for quote verification and traceability.

The Claude Code story mining pipeline generates both human-readable and machine-readable file outputs. It produces stories.md and digest.md for reading, alongside index.json, all written to the default .story-miner directory with full provenance tracking.