Capture lessons from Claude Code sessions into workspace memory.
npx skills add https://github.com/MDGrey33/memnyx --skill lessons-mdgrey33
Please help me install this Agent Skill. Skill: lessons Source: https://github.com/MDGrey33/memnyx/tree/main/.claude/skills/lessons Command: npx skills add https://github.com/MDGrey33/memnyx --skill lessons-mdgrey33
This Skill turns session takeaways into durable memory so recurring mistakes, user preferences, and workflow patterns are not lost between Claude Code sessions.
Ask the lessons skill to capture one lesson or scan recent sessions for skill-change proposals.
High-intent search queries and answers about installing and using this skill.
To save lessons from a Claude Code session, you can use the lessons skill to capture individual takeaways or scan session summaries directly into workspace memory files with proper categorization.
You can perform deep transcript analysis on JSONL files to mine for stronger signals like retry loops, tool failures, and silent skill breakdowns, turning them into structured lessons.
The best way to generate skill improvement proposals is to scan recent session summaries or transcripts, draft structured change proposals, and hand them off to a skills-manager instead of editing files directly.
Capturing session lessons requires workspace scope resolution to correctly target the memory file, and it supports optional cognee storage for persisting the categorized lessons.
Captured lessons are categorized during the capture process to ensure recurring mistakes, user preferences, and workflow patterns are organized and retained in workspace memory across sessions.
Use deep transcript scans instead of default session summary scans when you need to mine JSONL transcripts for stronger behavioral signals like retry loops, frustration, and silent skill breakdowns.
Largo2z9Export Claude Code sessions into structured markdown transcripts for archiving.
SanctifiedOpsModel pump.fun bonding curves, graduation triggers, and Raydium migration dynamics.
bykoleksii-hardoIntegrate persistent semantic memory with vector stores and knowledge graphs.
humanerd-drewCategorize skills as dead or active using cron registration, wikilinks, and mtime thresholds.
