Extract session patterns from JSONL metrics into structured learnings.
npx skills add https://github.com/Kanevry/session-orchestrator --skill evolve-kanevry
Please help me install this Agent Skill. Skill: evolve Source: https://github.com/Kanevry/session-orchestrator/tree/main/skills/evolve Command: npx skills add https://github.com/Kanevry/session-orchestrator --skill evolve-kanevry
Evolve extracts recurring patterns and quality signals from interactive session history and converts them into explicit, managed learnings to improve future session planning and execution.
Run the evolve skill in analyze mode to extract patterns from your session history and confirm which learnings to save.
High-intent search queries and answers about installing and using this skill.
You can extract recurring issues and fragile file patterns from session metrics by running an analyze process on historical session JSONL files to identify deviation patterns, sizing insights, and stagnation-class frequency across waves. This surfaces files that frequently break for targeted review.
The best way to deduplicate and persist learnings is to use atomic writes to update a learnings.jsonl file, which prevents data corruption. The process updates confidence scores, sets expirations, and requires user confirmation before saving extracted session patterns.
Yes, you can mirror confirmed learnings to an Obsidian vault using a configurable mirror script. This vault integration supports warn and strict modes to safely synchronize high-confidence guidance into your knowledge base.
To identify scope guidance and deviation patterns during project orchestration, apply pattern extraction to session metrics. This analyzes historical waves to detect stagnation-class frequency and recurring issues, converting them into structured learnings for future planning.
Yes, user confirmation is required before performing atomic rewrites to the learnings.jsonl file. This safeguard ensures you review and approve extracted session patterns, fragile files, and scope guidance before they are persisted or mirrored.
lemonlqfSearch and analyze JSONL conversation logs using jq and rg.
MeshJSStore and retrieve session learnings across Claude Code sessions in .claude/journeys/ files.
eterdisScan six PESTEL categories and update a watch list with alerts.
arbisoftObserve Claude Code sessions and convert recurring behaviors into atomic instincts with confidence scoring.
