learned-skills-index
Index directory for automatically learned skills from execution feedback
When & Why to Use This Skill
The Learned Skills Index is a sophisticated self-improvement framework for Claude agents that captures, validates, and indexes successful execution patterns. By transforming feedback from iterative tasks into a persistent knowledge base, it enables the agent to 'remember' and reuse high-quality solutions, significantly boosting efficiency and accuracy in complex, recurring workflows.
Use Cases
- Autonomous Pattern Optimization: Automatically extracting successful coding or logic patterns from execution feedback to improve the agent's performance over time.
- Context-Aware Skill Retrieval: Dynamically loading domain-specific expertise (e.g., specialized backend authentication or frontend form handling) based on the current task's requirements.
- Quality-Controlled Automation: Implementing a rigorous promotion gate that ensures only skills meeting high-performance thresholds (85%+ quality) are reused in production environments.
- Behavioral Auditing and Safety: Utilizing full provenance tracking to monitor, audit, or rollback learned behaviors, ensuring agentic actions remain transparent and safe.
| name | learned-skills-index |
|---|---|
| description | Index directory for automatically learned skills from execution feedback |
| type | index |
Learned Skills Index
This directory contains skills that have been automatically extracted from successful execution sessions. These skills represent patterns that led to quality improvements and are available for retrieval in future tasks.
How Skills Are Learned
- Execution: Claude Code runs a task with
--loopenabled - Feedback Collection: Each iteration's quality scores and improvements are recorded
- Pattern Extraction: Successful patterns are extracted from quality-improving iterations
- Skill Generation: Patterns are compiled into a learnable skill definition
- Promotion Gate: Skills must meet quality thresholds before promotion
Skill Status
| Status | Meaning |
|---|---|
| Pending | Skill extracted but not yet promoted (needs more validation) |
| Promoted | Skill has been validated and can be applied to new tasks |
| Archived | Skill deprecated or superseded by newer learning |
Quality Thresholds for Promotion
- Minimum quality score: 85.0
- Minimum successful applications: 2
- Minimum success rate: 70%
Directory Structure
learned/
├── SKILL.md # This index file
├── learned-backend-auth/ # Example learned skill
│ ├── SKILL.md # Skill definition
│ └── metadata.json # Machine-readable metadata
└── learned-frontend-form/ # Another learned skill
├── SKILL.md
└── metadata.json
Integration
Learned skills are automatically retrieved based on:
- Task description keywords
- File types being modified
- Domain context
To manually query learned skills:
from core.skill_persistence import retrieve_skills_for_task
skills = retrieve_skills_for_task(
task_description="your task here",
domain="backend" # optional
)
Safety
All learned skills include full provenance:
- Source session ID
- Source repository
- Timestamp of learning
- Quality progression
This enables rollback and audit of any behavioral changes from learned skills.