self-improvement

Captures learnings, errors, and corrections to enable continuous improvement. Use when: (1) A command or operation fails unexpectedly, (2) User corrects Claude ('No, that's wrong...', 'Actually...'), (3) User requests a capability that doesn't exist, (4) An external API or tool fails, (5) Claude realizes its knowledge is outdated or incorrect, (6) A better approach is discovered for a recurring task. Also review learnings before major tasks.

Quality

80%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

Securityby

Critical

Do not install without reviewing

Fix and improve this skill with Tessl

tessl review fix ./configs/microservice/bff-service/configs/agent-skills/clawhub/self-improvement/SKILL.md

Quality

Content

77%

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

A highly actionable, well-sequenced skill with concrete commands, templates, and working scripts, plus clear validation checkpoints. Its drawbacks are verbosity from repeated sections and an orphaned reference file that weakens progressive disclosure.

Suggestions

Collapse the three duplicated Promotion Targets tables and the overlapping OpenClaw/Generic setup sections into a single canonical version, referencing details once to cut the body substantially.

Link references/examples.md from the body (e.g., in the Logging Format or Automatic Skill Extraction sections) so it is discoverable, or remove it if unused.

Move the three full entry-format templates and the OpenClaw workspace structure into references/ files, keeping only one compact example inline in SKILL.md.

Dimension	Reasoning	Score
Conciseness	The body is procedural and accurate rather than explaining concepts Claude already knows, but at ~640 lines it carries notable redundancy — the Promotion Targets table appears three times, the OpenClaw and Generic setup paths overlap, and Detection Triggers repeats the Quick Reference — so it could be tightened considerably.	2 / 3
Actionability	It provides executable bash (mkdir/printf/grep), copy-paste-ready entry-format templates, concrete hook JSON configs, and real, functional scripts (the extraction helper includes argument parsing, validation, and dry-run), all directly usable.	3 / 3
Workflow Clarity	Multi-step processes are clearly numbered (first-use init, resolving, promoting, simplify-and-harden ingestion, extraction) with explicit validation such as idempotency checks, search-before-create, and a dry-run, plus checklists like the Skill Quality Gates.	3 / 3
Progressive Disclosure	References are one level deep and mostly well-signaled ("See references/openclaw-integration.md for complete details"), but references/examples.md is never linked from the body and the monolithic length includes inline content (workspace structure, three full entry templates, duplicated tables) that could be split out.	2 / 3
	Total	10 / 12 Passed

Description

82%

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, third-person description with explicit what-and-when guidance and natural, quoted trigger phrases. Its main weakness is a single-verb action statement and a few overly broad triggers that could invite overlap with error-handling skills.

Suggestions

Replace the single verb with multiple distinct actions (e.g., "Logs learnings, errors, and feature requests to markdown files, links related entries, and promotes high-value learnings to project memory") to raise specificity.

Tighten the broad failure triggers so they point specifically at learning capture rather than general error handling (e.g., "After resolving an unexpected command or API failure, log what went wrong and the fix").

Dimension	Reasoning	Score
Specificity	It names the domain and concrete capture targets ("Captures learnings, errors, and corrections") but relies on a single verb applied to three object types rather than listing multiple distinct actions, so it falls short of the multi-action anchor for 3.	2 / 3
Completeness	It explicitly states what the skill does and provides an explicit "Use when:" clause with six triggers plus "review learnings before major tasks," satisfying both what and when.	3 / 3
Trigger Term Quality	It quotes natural user phrases ("No, that's wrong...", "Actually...") and gives six numbered, conversational triggers, covering the kinds of statements users actually make.	3 / 3
Distinctiveness Conflict Risk	The correction and missing-capability triggers are distinctive, but "command or operation fails unexpectedly" and "external API or tool fails" are broad enough to overlap with general error-handling or debugging skills.	2 / 3
	Total	10 / 12 Passed

Validation

93%

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 — 15 / 16 Passed

Validation for skill structure

Criteria	Description	Result
skill_md_line_count	SKILL.md is long (645 lines); consider splitting into references/ and linking	Warning

	Total	15 / 16 Passed

Repository: UnicomAI/wanwu
Path: configs/microservice/bff-service/configs/agent-skills/clawhub/self-improvement/SKILL.md
Commit: f618458

Reviewed: about 13 hours 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.