CtrlK
BlogDocsLog inGet started
Tessl Logo
SkillPublic

self-healing-ci

CI-only self-healing workflow using gh-aw (GitHub Agentic Workflows) for active runtime recovery on pull requests and scheduled runs. When a CI check fails (test, build, lint, deploy, scan), this skill diagnoses the failure from CI logs, proposes a verified patch as a PR comment or follow-up commit, and commits a HEAL entry to `.learnings/HEALS.md`. Verify-before-persist discipline preserved: a HEAL is only `verified` if a re-run check passes in the same workflow; otherwise it ships as `pending-verify` for human follow-up. Recurrent heal patterns across PRs accumulate `Recurrence-Count` and append a `Handoff` block at ≥3 to flag promotion via self-improvement-ci. Use this skill when: you want headless heal-loop execution in CI/scheduled pipelines, you want recurring failure patterns captured automatically, or you want PRs that surface non-obvious environmental / tooling fixes without human triage. For interactive/local sessions, use `self-healing` instead.

Quality

88%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide
SecuritybySnyk

Low

Low-risk findings worth noting

SKILL.md
Quality
Evals
Security

Quality

Content

85%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, operationally concrete skill body: verification discipline is the explicit spine, authoring detail is properly disclosed into a real one-level-deep reference, and anti-patterns address real CI hazards. The only notable gap is that the HEAL entry file format is delegated to the sibling skill rather than shown or excerpted here.

Suggestions

Include a minimal HEAL entry template (or a link to the exact section of the sibling skill / reference that defines it) so a CI agent reading only this skill can file correctly formatted entries.

DimensionReasoningScore

Conciseness

The body is dense and operational (install commands, contract list, output schema, two tables, anti-pattern list), with over-explanation confined to the "Context Limitation (Important)" section and the "Never fake verified" rationale paragraph. Matches anchor 4 — efficient with minor instances that could be trimmed — rather than 5, and is well above anchor 3's "several unnecessary explanations".

4 / 5

Actionability

Concrete guidance throughout: executable install commands (`gh extension install github/gh-aw`), a complete machine-readable YAML output schema, real trigger events (`workflow_run`, `schedule`, `workflow_dispatch`), and a specific loop-guard (`if: github.actor != 'github-actions[bot]'`). Not anchor 5 because the HEAL entry markdown format itself is never shown — it is only deferred to the sibling skill.

4 / 5

Workflow Clarity

The verify-before-persist loop is fully operationalized: a stated diagnose → patch → verify → file sequence, a numbered CI contract, a failure-type → re-run verification table, an explicit `pending-verify` fallback when re-runs are infeasible, and labeled handling of abandoned heals. This matches anchor 5's explicit validation steps with error-recovery paths; the destructive/batch cap does not apply since validation is the skill's spine.

5 / 5

Progressive Disclosure

Well-organized sections with workflow-authoring detail correctly split into `references/workflow-example.md` — clearly signaled, one level deep, and verified to contain the promised workflow templates and checklists. Cross-references to sibling skills are explicit and navigable, matching anchor 5.

5 / 5

Total

18

/

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 explicitly covers what, when, and when-not with a clear CI-only niche and concrete capability verbs. Its weaknesses are mild: some token budget goes to internal state mechanics, and a few natural trigger phrasings users would actually say are missing.

Suggestions

Trim the internal-mechanics sentences (Recurrence-Count accumulation, Handoff block details) from the description and keep them in the body — they describe state, not user-facing capabilities.

Add one or two natural trigger phrasings users would actually say, e.g. "fix a failing CI check" or "red build on a PR", and replace the jargon phrase "headless heal-loop execution".

Rewrite the trigger clause in third person ("Use when CI checks fail on pull requests or scheduled runs") to match the voice of the capability statements.

DimensionReasoningScore

Specificity

Lists several concrete actions ("diagnoses the failure from CI logs", "proposes a verified patch as a PR comment or follow-up commit", "commits a HEAL entry to .learnings/HEALS.md") with only minor coverage gaps. Falls short of anchor 5 because a sizable share of the description covers internal bookkeeping mechanics (Recurrence-Count, Handoff blocks) rather than user-facing capabilities, and edges into second-person trigger phrasing ("you want headless heal-loop execution").

4 / 5

Completeness

Explicitly answers both what (diagnose, propose patch, file HEAL, track recurrence/promotion) and when ("Use this skill when: ..." with three concrete triggers), plus a negative boundary ("For interactive/local sessions, use `self-healing` instead"). Clearly matches the anchor-5 example structure.

5 / 5

Trigger Term Quality

Natural phrases like "When a CI check fails (test, build, lint, deploy, scan)", "pull requests and scheduled runs", and "CI/scheduled pipelines" give good keyword coverage. Not anchor 5 because "headless heal-loop execution" is author jargon and common user phrasings ("fix the red build", "flaky CI", "CI is broken") are absent.

4 / 5

Distinctiveness Conflict Risk

"CI-only" scoping plus the explicit handoff of interactive/local sessions to the sibling skill carves a clear niche with distinct triggers. Downstream overlap with self-improvement-ci is explicitly framed as a consumer relationship, so conflict risk is minimal.

5 / 5

Total

18

/

20

Passed

Validation

93%

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

CriteriaDescriptionResult

relative_links

Relative link issues: 6 suspicious

Warning

Total

15

/

16

Passed

Repository

pskoett/pskoett-ai-skills

Path

skills/self-healing-ci/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.