Session Learning

This extracts insights from your coding sessions and surfaces them when relevant later. At session end, it scans the transcript for error patterns, workflow discoveries, and debugging strategies, then stores them in categorized YAML files (errors, workflows, tools, architecture, debugging). At session start, it uses keyword matching to inject past learnings that relate to your current task. The matching is deliberately simple, no ML complexity, just keyword overlap weighted by confidence and recency. Think of it as institutional memory that complements your DISCOVERIES.md with structured, searchable learnings. Most useful when you're repeatedly hitting similar problems or want to remember how you solved something three weeks ago.

skills.sh installs: 151
GitHub stars: 76
last updated: Aug 21, 2026

View on GitHub

Install

$ npx -y skills add rysweet/amplihack --skill session-learning --agent claude-code

Installs into .claude/skills of the current project.

Files

Session Learning Skill

Purpose

This skill provides cross-session learning by:

Extracting learnings from session transcripts at Stop hook
Storing learnings in structured YAML format (~/.amplihack/.claude/data/learnings/)
Injecting relevant past learnings at SessionStart based on task similarity
Managing learnings via /amplihack:learnings capability