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wrong...', 'Actually...'), (2) User requests a capability that doesn't exist, (3) Claude realizes its knowledge is outdated or incorrect, (4) A better approach is discovered for a recurring task, (5) Receiving a Handoff block from self-healing (a recurring verified heal with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

Quality

80%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide
SecuritybySnyk

Passed

No findings from the security scan

Fix and improve this skill with Tessl

tessl review fix ./plugin/skills/self-improvement/SKILL.md

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

SKILL.md
Quality
Evals
Security

Quality

Content

67%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, highly actionable instruction skill: concrete entry templates, executable commands, numbered workflows with human-approval and dry-run checkpoints, and proper deferral of setup detail to references. Main weaknesses are noticeable token padding in taxonomy/best-practice/peripheral sections, a nested-fence defect in the Error Entry template, and an orphaned examples.md reference.

Suggestions

Trim or relocate sections that restate knowledge Claude already has — Priority Guidelines, Area Tags, Best Practices, Gitignore Options, and Copilot Chat Integration — into a reference file or drop them, keeping the body focused on entry formats, the promotion rule, and workflows.

Fix the nested triple-backtick fence in the Error Entry template (the inner ``` block terminates the outer ```markdown fence); use a four-backtick outer fence or indent the error snippet so the template renders and copies correctly.

Link references/examples.md from the body (e.g., from the Logging Format section: 'for fully worked entry examples, see references/examples.md') so the 383-line examples file is discoverable instead of orphaned.

DimensionReasoningScore

Conciseness

The core content (entry templates, ID format, promotion rule, ingestion workflow) is domain-specific and efficient, but several sections restate knowledge Claude already has: the Priority Guidelines and Area Tags tables define a taxonomy Claude could infer, 'Best Practices' items like 'Log immediately' and 'Be specific' are generic filler, and Gitignore Options, Multi-Agent Support, and Copilot Chat Integration add padding. More than minor over-explanation (below anchor 4) but the body is not dominated by it (above anchor 2).

3 / 5

Actionability

Provides copy-paste-ready markdown templates for all three entry types, an exact ID format with examples, executable commands (mkdir -p .learnings, grep one-liners for status checks, extract-skill.sh with --dry-run), and concrete trigger phrases to detect. Minor gaps prevent a 5: the Error Entry template nests triple-backtick fences inside a triple-backtick block (breaking rendering and copy-paste fidelity), and the extract-skill.sh invocation hardcodes a repo-specific path ('./skills/self-improvement/scripts/...') that differs from the frontmatter's CLAUDE_PLUGIN_ROOT layout.

4 / 5

Workflow Clarity

Multi-step workflows are clearly numbered: promotion has a 5-step sequence with a human-approval checkpoint ('Show the proposed diff for human approval before writing instructions'), extraction runs --dry-run first and includes a quality-gates checklist plus a 'Verify: read skill in fresh session' step, and recurring-pattern detection specifies search-first then link/bump. Below anchor 5 because some flows lack validation (appending an entry has no format check, periodic review is described loosely) — these are minor gaps consistent with anchor 4 rather than the explicit validate-fix-retry loops of anchor 5.

4 / 5

Progressive Disclosure

Good bundle discipline: installation, hooks, Amp setup, and context-maintenance detail are correctly deferred to one-level-deep reference files with explicit read-when guidance ('Read these only when installing or configuring the skill'), and templates live in assets/. Minor gaps: references/examples.md (383 lines) is never linked from the body, making it undiscoverable, and peripheral sections (Gitignore Options, Copilot Chat Integration) would fit better in a reference file — consistent with anchor 4's 'minor organization gaps' rather than anchor 5's fully clean split.

4 / 5

Total

15

/

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: it states what the skill does, gives five explicit numbered trigger conditions with natural user phrasing, and proactively disambiguates the two nearest competing skills (self-healing, self-improvement-ci). The only softness is that actions are described at a process level without the concrete artifacts the skill actually produces (.learnings/ entry files).

DimensionReasoningScore

Specificity

Lists several concrete actions — 'Captures learnings, errors, corrections, and feature requests', 'distill into a memory file or new skill', 'promotes accumulated patterns', 'review learnings before major tasks' — covering the skill's main behaviors. Not 5 because the actions are process-level verbs without concrete artifacts (no mention of the .learnings/ files or entry formats it actually writes); not 3 because coverage goes well beyond 1-2 actions.

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 of five concrete trigger scenarios, plus boundary disambiguation for self-healing and self-improvement-ci). Both what and when are stated with concrete trigger phrases, matching the anchor-5 example structure exactly.

5 / 5

Trigger Term Quality

Includes genuinely natural user utterances as triggers — 'No, that's wrong...', 'Actually...', 'User requests a capability that doesn't exist', 'knowledge is outdated or incorrect', 'better approach is discovered'. A few common natural terms are missing (e.g., 'remember this', 'save this learning', 'learn from mistakes'), which keeps it below the comprehensive synonym coverage of anchor 5, but coverage is clearly above the partial coverage of anchor 3.

4 / 5

Distinctiveness Conflict Risk

Occupies a clear niche (passive capture/promotion of learnings) and explicitly delineates the adjacent skills: '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 active boundary guidance minimizes wrong-skill triggering, matching the clear-niche anchor.

5 / 5

Total

18

/

20

Passed

Validation

81%

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

Validation — 13 / 16 Passed

Validation for skill structure

CriteriaDescriptionResult

skill_md_line_count

SKILL.md is long (518 lines); consider splitting into references/ and linking

Warning

frontmatter_unknown_keys

Unknown frontmatter key(s) found; consider removing or moving to metadata

Warning

relative_links

Relative link issues: 1 suspicious

Warning

Total

13

/

16

Passed

Repository

pskoett/pskoett-ai-skills

Path

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.