self-improvement

Captures learnings, errors, corrections, and feature requests to enable continuous improvement. Use when: (1) User corrects Claude ('No, that's wrong...', 'Actually...'), (2) User requests a capability that doesn't exist, (3) Claude realizes its knowledge is outdated or incorrect, (4) A better approach is discovered for a recurring task, (5) Receiving a Handoff block from self-healing (a recurring verified heal with Recurrence-Count at least 3) to distill into a memory file or new skill. For ACTIVE runtime failures where the agent needs to apply and verify a fix mid-task, use `self-healing` instead (it files HEAL- entries with proof; self-improvement promotes accumulated patterns). Also review learnings before major tasks. For CI-only/headless learning capture, use self-improvement-ci.

4.40x

Quality

85%

Does it follow best practices?

Impact

97%

4.40x

Average score across 3 eval scenarios

Securityby

Low

Low-risk findings worth noting

Quality

Content

75%Weight 40%Scale 1-5

Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.

Highly actionable body with concrete templates, commands, and hook configs, plus well-sequenced workflows with verification for skill extraction. The main weaknesses are redundancy of trigger phrasing across sections and an unreferenced examples.md bundle file.

Suggestions

Deduplicate trigger phrases: keep them in Detection Triggers and have the Quick Reference table point there instead of restating the same 'No, that's wrong...' / 'Actually...' cues.

Link references/examples.md from the body (e.g., under Logging Format or Best Practices) so the bundled examples file is discoverable, or remove it if unused.

Add an explicit post-write validation checkpoint to the 'How to Promote' workflow (e.g., re-read the inserted rule in CLAUDE.md/AGENTS.md context to confirm it reads correctly) to match the verification discipline already present in skill extraction.

Dimension	Reasoning	Score
Conciseness	Mostly actionable reference content, but the trigger phrases recur in three places (description, Quick Reference, Detection Triggers) and the Multi-Agent section repeats hook details already in Hook Integration, so it could be tightened by deduplicating.	3 / 5
Actionability	Provides copy-paste-ready markdown entry templates, exact ID-generation formats, concrete bash/grep commands, full hook JSON configs with real paths, and explicit script invocations, fully covering the common logging and promotion cases.	5 / 5
Workflow Clarity	Promotion, recurring-pattern ingestion, and skill-extraction workflows are clearly numbered, and extraction includes a --dry-run plus a 'Verify in fresh session' checkpoint; minor gap is that promotion to shared context files lacks an explicit post-write validation step.	4 / 5
Progressive Disclosure	Clear sectioned overview with one-level-deep, well-signaled references (openclaw-integration.md, hooks-setup.md, assets/templates) all of which exist, but references/examples.md is bundled yet never linked from the body and some inline agent-setup content could arguably live in a reference.	4 / 5
	Total	16 / 20 Passed

Description

95%Weight 40%Scale 1-5

Based on the skill's description, can an agent find and select it at the right time? Clear, specific descriptions lead to better discovery.

A strong, well-structured description that clearly states what the skill does and gives five explicit 'Use when' triggers plus disambiguation from self-healing and self-improvement-ci. Slightly less granular in its action verbs than the top specificity anchor, but otherwise excellent.

Dimension	Reasoning	Score
Specificity	Lists several concrete capture types ('learnings, errors, corrections, and feature requests') plus 'distill into a memory file or new skill' and 'promotes accumulated patterns', but the verbs are somewhat repetitive and the actions are less granular than the 5-anchor's distinct operations.	4 / 5
Completeness	Explicitly answers both what ('Captures learnings, errors, corrections, and feature requests to enable continuous improvement') and when (a numbered 'Use when:' list with five concrete trigger conditions), matching the 5-anchor.	5 / 5
Trigger Term Quality	Embeds natural user phrases users actually say ('No, that's wrong...', 'Actually...') alongside scenario triggers like requesting a missing capability and discovering outdated knowledge, giving comprehensive coverage of natural terms.	5 / 5
Distinctiveness Conflict Risk	Explicitly disambiguates from sibling skills ('use `self-healing` instead' for active runtime failures; 'use self-improvement-ci' for CI/headless capture), carving a clear niche with minimal conflict risk.	5 / 5
	Total	19 / 20 Passed

Validation

87%

Warnings & errors only

Checks the skill against the spec for correct structure and formatting. All validation checks must pass before discovery and implementation can be scored.

Validation — 14 / 16 Passed

Validation for skill structure

Criteria	Description	Result
skill_md_line_count	SKILL.md is long (579 lines); consider splitting into references/ and linking	Warning
relative_links	Relative link issues: 1 suspicious	Warning

	Total	14 / 16 Passed

Repository: pskoett/pskoett-ai-skills
Path: skills/self-improvement/SKILL.md
Commit: 8a71d70

Reviewed: 12 days ago

Table of Contents

Discovery Implementation Validation

Is this your skill?

If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.