CtrlK
BlogDocsLog inGet started
Tessl Logo
SkillPublic

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wrong...', 'Actually...'), (2) User requests a capability that doesn't exist, (3) Claude realizes its knowledge is outdated or incorrect, (4) A better approach is discovered for a recurring task, (5) Receiving a Handoff block from self-healing (a recurring verified heal with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.

0.89x
Quality

83%

Does it follow best practices?

Impact

84%

0.89x

1 of 3 eval scenarios. Add 2 more for a full score.

SecuritybySnyk

Passed

No findings from the security scan

The canonical home for this skill is self-improvement in pskoett/pskoett-ai-skills

SKILL.md
Quality
Evals
Security

Quality

Content

73%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.

A well-structured operational skill with copy-paste entry templates, exact commands, and exceptionally well-sequenced workflows including human-approval gates, dry-runs, and verification steps. The main weaknesses are moderate verbosity (duplicated trigger lists and tables explaining concepts Claude already knows), a broken code fence inside the error-entry template, and an orphaned examples.md reference.

Suggestions

Consolidate the three overlapping trigger lists (Quick Reference, 'Detection Triggers', 'Multi-Agent Support') into the Quick Reference table, and drop or trim the 'Area Tags' and 'Priority Guidelines' tables — Claude can infer what 'frontend' and 'critical' mean.

Fix the nested code fence in the Error Entry template (use a four-backtick outer fence or an indented block for the '### Error' section) so the template renders and copies as one unit.

Link references/examples.md from the 'Logging Format' section (e.g., 'For fully worked entries, see references/examples.md') so the largest reference file is discoverable.

DimensionReasoningScore

Conciseness

The core is efficient (copy-paste entry templates, ID format, exact grep commands), but several sections pad or duplicate: 'Area Tags' explains that frontend means 'UI, components, client-side code', the 'Priority Guidelines' table spells out obvious severity levels, and trigger guidance appears three times (Quick Reference, 'Detection Triggers', and 'Multi-Agent Support'). Matches anchor 3 ('mostly efficient but includes some unnecessary explanation or could be tightened'); not 2 because the bulk of the body is genuinely operational rather than padded.

3 / 5

Actionability

Highly executable throughout: full markdown entry templates, 'grep -r "keyword" .learnings/', './skills/self-improvement/scripts/extract-skill.sh skill-name --dry-run', and a numbered promotion workflow with an explicit human-approval checkpoint. Falls short of 5 on two minor gaps: the error-entry template's nested triple-backtick fence is broken in the source (the inner ``` at the '### Error' block prematurely closes the outer ```markdown fence, so lines like 'Actual error message or output' render outside the template), and the extract-skill.sh path hardcodes one install layout.

4 / 5

Workflow Clarity

Multi-step processes are explicitly sequenced with validation checkpoints: promotion requires 'Show the proposed diff for human approval before writing instructions' and 'Do not mark an unapplied proposal as promoted'; skill extraction runs '--dry-run' before the real run and ends with 'Verify: Read skill in fresh session to ensure it's self-contained'; there is a quality-gates checklist and a recurrence feedback loop (increment Recurrence-Count, bump priority, consider systemic fix). Matches anchor 5; anchor 4 would imply missing checkpoints, which is not the case.

5 / 5

Progressive Disclosure

Good one-level-deep structure: 'For installation commands, see references/installation.md... Read these only when installing or configuring the skill', with hooks-setup.md, amp-setup.md, context-maintenance.md, assets/ templates, and scripts/extract-skill.sh all correctly signposted and all verified to exist in the bundle. Falls short of 5 because references/examples.md — the largest reference file (383 lines of worked entry examples) — is never linked from SKILL.md anywhere, leaving it undiscoverable.

4 / 5

Total

16

/

20

Passed

Description

87%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 description that clearly states what the skill does, gives five numbered concrete trigger conditions with natural user phrasing, and explicitly disambiguates against the adjacent self-healing and self-improvement-ci skills. Minor room to add a few more natural trigger synonyms and an operational detail about where entries are written.

DimensionReasoningScore

Specificity

Quotes several concrete actions — 'Captures learnings, errors, corrections, and feature requests', 'distill into a memory file or new skill', 'review learnings before major tasks' — which match anchor 4 ('lists several specific actions; minor gaps in coverage'). Not 5 because operational details like where entries are written (.learnings/ files) and promotion targets are absent; not 3 because well more than 1-2 actions are named.

4 / 5

Completeness

Explicitly answers both: what ('Captures learnings, errors, corrections, and feature requests to enable continuous improvement') and when ('Use when: (1) User corrects Claude... (2) User requests a capability that doesn't exist... (3) Claude realizes its knowledge is outdated...'). Matches anchor 5 with concrete numbered trigger phrases; a 4 would require the 'when' to be less explicit, which it is not.

5 / 5

Trigger Term Quality

Includes natural trigger phrases users would actually say: 'No, that's wrong...', 'Actually...', 'User requests a capability that doesn't exist', 'knowledge is outdated'. Good keyword coverage matching anchor 4; not 5 because common variations like 'remember this', 'save this learning', or 'learn from mistakes' are missing.

4 / 5

Distinctiveness Conflict Risk

Clear niche with active disambiguation: 'For ACTIVE runtime failures where the agent needs to apply and verify a fix mid-task, use self-healing instead' and 'For CI-only/headless learning capture, use self-improvement-ci'. This is anchor 5 — distinct triggers, minimal conflict risk; the neighboring anchor 4 would imply unresolved overlap, but the description resolves its overlaps explicitly.

5 / 5

Total

18

/

20

Passed

Validation

87%

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

CriteriaDescriptionResult

skill_md_line_count

SKILL.md is long (508 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

agent-plugin/skills/self-improvement/SKILL.md

Reviewed

Table of Contents

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.