skills.rest
lettucebo
lettuceboCOMMUNITY
Explore all skills by this author →
WebsiteGitHub

continual-learning

Implements persistent memory and learning loops for AI coding agents across sessions.

Updated Mar 25, 2026
Search Tags:#hooks#ai agent#memory#sqlite#reflection#continual learning#copilot
One-click install
npx skills add https://github.com/lettucebo/Skills --skill continual-learning-lettucebo
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: continual-learning
Source: https://github.com/lettucebo/Skills/tree/main/skills/microsoft/continual-learning
Command: npx skills add https://github.com/lettucebo/Skills --skill continual-learning-lettucebo

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? AI coding agents forget everything between sessions, forcing users to repeat corrections and re-explain project conventions. This Skill sets up continual learning infrastructure so agents capture, persist, and apply learnings automatically. ## Core Features & Use Cases - Two-Tier Memory: Global learnings (tool patterns, cross-project conventions) stored in ~/.copilot/learnings.db, and local project-specific learnings stored in .copilot-memory/learnings.db. - Automatic Capture via Hooks: A hook observes tool outcomes, detects failure patterns, and surfaces relevant learnings at the start of the next session. - Agent-Native and Manual Storage: Agents can write learnings directly via SQL (categories: pattern, mistake, preference, tool_insight), while humans can maintain version-controlled markdown memory files. - Use Case: After your agent repeatedly fails with a bash command in one session, the hook stores that pattern; in the next session the learning is surfaced so the agent adjusts its approach without being told again. ## Quick Start Ask the agent to install the continual-learning hook by copying hooks/continual-learning into .github/hooks/ so it auto-initializes on the first session.

Frequently Asked Questions about continual-learning

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

FAQPage Schema
How do I make an AI coding agent remember things between sessions?▼

Install the continual-learning hook by copying hooks/continual-learning into .github/hooks/. It auto-initializes on the first session, observes tool outcomes, and stores learnings in a SQLite database that is surfaced at the start of later sessions.

What is the difference between global and local agent memory?▼

Global memory lives in ~/.copilot/learnings.db and follows you across all projects, storing tool patterns and general preferences. Local memory lives in .copilot-memory/learnings.db and stays with the repository for project-specific conventions and mistakes.

How do I store a learning manually for an AI agent?▼

Insert a row into the learnings table with a scope, category, content, and source, for example a local convention like using Result<T> instead of exceptions. Categories include pattern, mistake, preference, and tool_insight.

Does continual learning memory grow unbounded over time?▼

No. Entries older than 60 days with low hit counts are pruned, tool logs are removed after 7 days, and frequently referenced high-value learnings persist indefinitely. This compaction prevents unbounded growth.

When should I use memory files instead of the learnings database?▼

Use markdown memory files like .copilot-memory/conventions.md for human-readable, version-controlled knowledge such as exact parameter names. Use the database for automatically captured patterns and agent-written learnings.

Related Skills

Explore More Software Engineering Skills→
iBz-04iBz-04
63

continual-learning

Capture agent experiences and persist learnings across sessions via databases and memory files.

Community
Advanced
ReidondReidond

learning-consolidator

Consolidate project learning logs into AI rules, skills, and conventions.

Community
Advanced
mdjubaerhossensabbirmdjubaerhossensabbir

agent-customization

Manage VS Code agent customization files with YAML frontmatter validation.

Community
Advanced
evolv3aievolv3ai
3

google-gemini-file-search

Create RAG systems with Google Gemini's managed File Search API.

Community
Advanced