Self Improvement
Log corrections and insights during agent sessions, track resolution status, and promote repeated learnings into CLAUDE.md rules or new reusable skills.
Overview
Self-improvement is a journey-wide agent skill that captures dev learnings, tracks resolution, and promotes repeat fixes into agent instructions or new skills—usable whenever a solo builder needs to compound agent memory
Install
npx skills add https://github.com/pskoett/pskoett-ai-skills --skill self-improvementWhat is this skill?
- Learning categories: correction, insight, knowledge_gap, best_practice across frontend, backend, infra, tests, docs, con
- Six statuses including resolved, wont_fix, promoted, and promoted_to_skill with defined meanings
- Skill extraction template with frontmatter name, description, and trigger-oriented SKILL.md scaffold
- Promotion targets: CLAUDE.md, AGENTS.md, or copilot-instructions.md for repo-level agent instructions
- Example LRN-style entries with Priority, Area, Summary, and Skill-Path when elevated to a skill
- 4 learning categories and 7 workspace areas defined in the log schema
- 6 resolution statuses including promoted and promoted_to_skill
Adoption & trust: 1.3k installs on skills.sh; 203 GitHub stars; 3/3 security scanners passed (skills.sh audits).
What problem does it solve?
Your agent keeps rediscovering the same mistakes because corrections live in chat history instead of structured learnings or reusable skills.
Who is it for?
Solo builders running long-horizon agent projects who want a disciplined loop from correction → resolved → promoted_to_skill.
Skip if: One-shot tasks with no expectation of reuse, or teams that already enforce learnings solely through formal RFCs without agent instruction files.
When should I use this skill?
After agent-assisted work when you spot a correction, insight, or gap that should be tracked, resolved, or promoted to instructions or a new skill.
What do I get? / Deliverables
You maintain a categorized learning log and can promote high-priority items into repo instructions or a new skill with documented Skill-Path and triggers.
- Structured learning log entries with status and resolution
- Promoted instruction snippets or new SKILL.md from template
Recommended Skills
Journey fit
Useful at every journey phase - explore requirements and options before committing to a direction.
Where it fits
Log an API auth correction as high-priority infra learning before promoting a dedicated integration skill.
Turn recurring review feedback into a best_practice entry slated for AGENTS.md promotion.
Mark a production incident fix resolved and extract a wont_fix vs promoted decision with Skill-Path for the next deploy.
Capture prototype-only shortcuts as knowledge_gap items so they do not leak into production skills unreviewed.
How it compares
Meta workflow for agent memory and skill extraction—not a linter, test runner, or automatic benchmark harness.
Common Questions / FAQ
Who is self-improvement for?
Solo and indie developers who use Claude Code, Cursor, or similar agents and want corrections to become durable repo rules or standalone skills.
When should I use self-improvement?
After any agent session where you fixed something worth remembering—during Build integrations, Ship review follow-ups, Operate incident fixes, or Validate prototype tweaks—whenever a pattern should not be re-learned from scratch.
Is self-improvement safe to install?
It describes logging and promotion workflows locally in your repo; review the Security Audits panel on this Prism page and avoid promoting secrets into CLAUDE.md or skills.
SKILL.md
READMESKILL.md - Self Improvement
# Learnings Corrections, insights, and knowledge gaps captured during development. **Categories**: correction | insight | knowledge_gap | best_practice **Areas**: frontend | backend | infra | tests | docs | config **Statuses**: pending | in_progress | resolved | wont_fix | promoted | promoted_to_skill ## Status Definitions | Status | Meaning | |--------|---------| | `pending` | Not yet addressed | | `in_progress` | Actively being worked on | | `resolved` | Issue fixed or knowledge integrated | | `wont_fix` | Decided not to address (reason in Resolution) | | `promoted` | Elevated to CLAUDE.md, AGENTS.md, or copilot-instructions.md | | `promoted_to_skill` | Extracted as a reusable skill | ## Skill Extraction Fields When a learning is promoted to a skill, add these fields: ```markdown **Status**: promoted_to_skill **Skill-Path**: skills/skill-name ``` Example: ```markdown ## [LRN-20250115-001] best_practice **Logged**: 2025-01-15T10:00:00Z **Priority**: high **Status**: promoted_to_skill **Skill-Path**: skills/docker-m1-fixes **Area**: infra ### Summary Docker build fails on Apple Silicon due to platform mismatch ... ``` --- # Skill Template Template for creating skills extracted from learnings. Copy and customize. --- ## SKILL.md Template ```markdown --- name: skill-name-here description: "Concise description of when and why to use this skill. Include trigger conditions." --- # Skill Name Brief introduction explaining the problem this skill solves and its origin. ## Quick Reference | Situation | Action | |-----------|--------| | [Trigger 1] | [Action 1] | | [Trigger 2] | [Action 2] | ## Background Why this knowledge matters. What problems it prevents. Context from the original learning. ## Solution ### Step-by-Step 1. First step with code or command 2. Second step 3. Verification step ### Code Example \`\`\`language // Example code demonstrating the solution \`\`\` ## Common Variations - **Variation A**: Description and how to handle - **Variation B**: Description and how to handle ## Gotchas - Warning or common mistake #1 - Warning or common mistake #2 ## Related - Link to related documentation - Link to related skill ## Source Extracted from learning entry. - **Learning ID**: LRN-YYYYMMDD-XXX - **Original Category**: correction | insight | knowledge_gap | best_practice - **Extraction Date**: YYYY-MM-DD ``` --- ## Minimal Template For simple skills that don't need all sections: ```markdown --- name: skill-name-here description: "What this skill does and when to use it." --- # Skill Name [Problem statement in one sentence] ## Solution [Direct solution with code/commands] ## Source - Learning ID: LRN-YYYYMMDD-XXX ``` --- ## Template with Scripts For skills that include executable helpers: ```markdown --- name: skill-name-here description: "What this skill does and when to use it." --- # Skill Name [Introduction] ## Quick Reference | Command | Purpose | |---------|---------| | `./scripts/helper.sh` | [What it does] | | `./scripts/validate.sh` | [What it does] | ## Usage ### Automated (Recommended) \`\`\`bash ./skills/skill-name/scripts/helper.sh [args] \`\`\` ### Manual Steps 1. Step one 2. Step two ## Scripts | Script | Description | |--------|-------------| | `scripts/helper.sh` | Main utility | | `scripts/validate.sh` | Validation checker | ## Source - Learning ID: LRN-YYYYMMDD-XXX ``` --- ## Naming Conventions - **Skill name**: lowercase, hyphens for spaces - Good: `docker-m1-fixes`, `api-timeout-patterns` - Bad: `Docker_M1_Fixes`, `APITimeoutPatterns` - **Description**: Start with action verb, mention trigger - Good: "Handles Docker build failures on Apple Silicon. Use when builds fail with platform mismatch." - Bad: "Docker stuff" - **Files**: - `SKILL.md` - Required, main documentation - `scripts/` - Optional, executable code - `references/` - Optional, detailed docs - `assets/` - Optional, templates --- ## Extraction Checklist Before cr