Continuous Learning

Extract reusable patterns from sessions and save them as knowledge

besync-labs Updated

File contents

Continuous Learning Skill

Purpose: Learn from development sessions to improve future assistance


Overview

This skill implements the PAAL (Perceive-Analyze-Adapt-Learn) cycle to extract reusable patterns from development sessions.


PAAL Cycle

1. Perceive

Monitor session for significant events:

  • User corrections
  • Repeated patterns
  • Error resolutions
  • Decision rationale

2. Analyze

Identify learning opportunities:

  • What went wrong?
  • What went right?
  • What patterns emerged?
  • What knowledge was missing?

3. Adapt

Apply learnings immediately:

  • Update mental models
  • Adjust approach
  • Incorporate feedback

4. Learn

Persist knowledge for future:

  • Document patterns in decisions/
  • Update knowledge base
  • Create reusable templates

Pattern Extraction Format

# Pattern: [Name]

## Context

When: [Situation]
Problem: [What problem this solves]

## Solution

[How to apply the pattern]

## Example

[Concrete example]

## Anti-patterns

[What NOT to do]

Integration

  • Runs at session end
  • Can be triggered manually with /learn
  • Outputs to decisions/ directory

besync-labs/antigravity-ai-kit/tree/main/.agent/skills/continuous-learning commit 12a2b45cd9

Frequently asked questions

npx skillmds@latest add besync-labs/continuous-learning