skills.rest
taipt1504
taipt1504COMMUNITY
Explore all skills by this author →
GitHub

continuous-learning

Persist AI session learnings from Claude memory and MCP knowledge graphs.

3|2|Updated Jan 16, 2026
Search Tags:#automation#learning#metadata#session#knowledge-graph#instincts
One-click install
npx skills add https://github.com/taipt1504/agent-skills --skill continuous-learning-taipt1504
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: continuous-learning
Source: https://github.com/taipt1504/agent-skills/tree/main/skills/continuous-learning
Command: npx skills add https://github.com/taipt1504/agent-skills --skill continuous-learning-taipt1504

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) and references (resource) components.

What problem does it solve?

Developers lose valuable session knowledge because Claude’s memory is volatile and project decisions are not persisted. This makes it hard to recall preferences, architecture decisions, or recurring bug patterns across sessions.

Core Features & Use Cases

  • Observable learning system that tracks Claude’s native memory and MCP knowledge graph, storing entities such as architecture decisions, bug patterns, anti‑patterns, and project conventions.
  • Meta commands (/meta learn status, /meta learn extract, /meta learn report, /meta evolve, /meta prune) let users query, extract, evolve, and clean up learned knowledge.
  • Instinct extraction script (scripts/extract-instincts.sh) analyzes session metrics and workflow state to surface frequent skill usage, error patterns, and workflow deviations.
  • Cross‑project persistence via .claude/knowledge/ fallback when MCP tools are unavailable.

Quick Start

Ask the continuous-learning skill to summarize what you have learned by saying “/meta learn status”.

Frequently Asked Questions about continuous-learning

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I persist Claude session knowledge and architecture decisions across projects?▼

To persist Claude session knowledge, you can use an observable learning system that tracks native memory and the MCP knowledge graph. It stores architecture decisions and bug patterns in a `.claude/knowledge/` directory for cross-project access.

What is the best way to extract recurring bug patterns and workflow deviations from AI development sessions?▼

The best way to extract recurring bug patterns is by running an instinct extraction script. This script analyzes session metrics and workflow state to surface frequent skill usage, error patterns, and workflow deviations automatically.

Can I store project conventions and anti-patterns when MCP tools are unavailable?▼

Yes, you can store project conventions and anti-patterns without MCP tools. The system uses a `.claude/knowledge/` directory fallback to ensure cross-project persistence of learned entities during development sessions.

How do I query and clean up learned knowledge entities from previous development sessions?▼

You can query and clean up learned knowledge entities using meta commands. Commands like `/meta learn status`, `/meta learn report`, and `/meta prune` let you extract, evolve, and remove stored architecture decisions and bug patterns.

Does capturing AI session learnings require external dependencies or complex setup?▼

Capturing AI session learnings requires no external dependencies. It operates directly on Claude memory and the MCP knowledge graph, needing only access to the `.claude/knowledge` directory and an optional shell script for pattern mining.

Related Skills

Explore More Software Engineering Skills→
rhpdsrhpds

03-memory-mcp

Configure persistent Memory MCP knowledge graph storage for Claude Code sessions.

Official
Advanced
jaaaackieLaijaaaackieLai

continuous-learning

Capture Claude Code session patterns into confidence-scored instincts for reuse.

Community
Advanced
yebotyebot
9

learning-taxonomy

Categorize session learnings into taxonomy types and route them to memory files.

Community
Advanced
chrono-metachrono-meta
7

field-harvest

Scan git commit history to detect reusable patterns and log session work.

Community
Advanced