self-improvement-ci
CI-only self-improvement workflow using gh-aw (GitHub Agentic Workflows). Captures recurring failure patterns and quality signals from pull request checks, emits structured learning candidates, and proposes durable prevention rules without interactive prompts. Use when: you want automated learning capture in CI/headless pipelines.
59
67%
Does it follow best practices?
Impact
—
No eval scenarios have been run
Passed
No known issues
Optimize this skill with Tessl
npx tessl skill review --optimize ./skills/self-improvement-ci/SKILL.mdQuality
Discovery
85%Based on the skill's description, can an agent find and select it at the right time? Clear, specific descriptions lead to better discovery.
This is a well-structured description that clearly defines a specific niche (CI-only self-improvement via gh-aw) with concrete actions and an explicit 'Use when' clause. Its main weakness is that trigger terms lean toward specialized jargon rather than natural language a user might type, which could reduce discoverability. The 'Use when' clause could also benefit from more varied trigger scenarios.
Suggestions
Add more natural trigger terms users might say, such as 'CI/CD failures', 'build errors', 'GitHub Actions', 'automated code quality', or 'pipeline debugging'.
| Dimension | Reasoning | Score |
|---|---|---|
Specificity | Lists multiple concrete actions: 'captures recurring failure patterns and quality signals from pull request checks', 'emits structured learning candidates', and 'proposes durable prevention rules without interactive prompts'. These are specific, actionable capabilities. | 3 / 3 |
Completeness | Clearly answers both 'what' (captures failure patterns, emits learning candidates, proposes prevention rules) and 'when' ('Use when: you want automated learning capture in CI/headless pipelines'). The explicit 'Use when' clause is present. | 3 / 3 |
Trigger Term Quality | Includes some relevant terms like 'CI', 'pull request checks', 'headless pipelines', 'gh-aw', and 'GitHub Agentic Workflows'. However, it uses fairly specialized jargon ('structured learning candidates', 'durable prevention rules') and misses common user terms like 'CI/CD', 'build failures', 'automated fixes', or 'GitHub Actions'. | 2 / 3 |
Distinctiveness Conflict Risk | Highly distinctive with a clear niche: CI-only self-improvement workflow using a specific tool (gh-aw). The combination of 'CI-only', 'headless', 'failure patterns', and 'prevention rules' makes it unlikely to conflict with other skills. | 3 / 3 |
Total | 11 / 12 Passed |
Implementation
50%Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
This skill provides a reasonable conceptual framework for CI-based self-improvement with a useful output schema and clear constraints, but falls short on actionability—the core operations (inspecting PR checks, ingesting candidates, deduplication) are described rather than demonstrated with executable code or commands. The authoring workflow section is the strongest part with concrete steps and validation, but the referenced template file is missing from the bundle.
Suggestions
Add executable code or specific CLI commands showing how to inspect PR check results and ingest learning candidates (e.g., `gh pr checks <PR> --json` or equivalent), rather than describing these steps abstractly.
Provide the referenced `references/workflow-example.md` template file in the bundle, or inline a minimal working example of the GitHub Actions workflow configuration.
Add an explicit numbered workflow for the core CI loop (inspect → ingest → deduplicate → emit) with validation checkpoints, similar to the authoring workflow section.
Trim the 'Context Limitation' section to a single bullet point—Claude can infer the implications from the constraint itself.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The skill is reasonably efficient but includes some sections that could be tightened—e.g., the 'Context Limitation' section explains reasoning Claude can infer, and the 'Purpose' section partially restates the description. The install block at the top adds tokens for a standard operation. However, it avoids egregious padding. | 2 / 3 |
Actionability | The skill provides some concrete commands (gh aw compile, gh aw run) and a clear YAML output schema, but the core workflow—how to actually inspect PR checks, ingest candidates, deduplicate by pattern_key, and emit output—is described abstractly rather than with executable code or specific commands. The authoring workflow references a template file that isn't provided in the bundle. | 2 / 3 |
Workflow Clarity | The authoring workflow has a clear 4-step sequence with a validation step (gh aw compile --validate --strict), which is good. However, the main CI workflow (inspect checks → ingest candidates → deduplicate → emit output) lacks explicit sequencing and has no validation/feedback loops for the core learning capture process. The CI contract lists constraints but not operational steps. | 2 / 3 |
Progressive Disclosure | The skill references 'references/workflow-example.md' for templates and mentions other skills for integration, which is good structure. However, no bundle files are provided, so the referenced template doesn't exist, and some content (like the full recurrence rules and promotion targets) could potentially be split out. The organization is decent but the missing reference file weakens the disclosure chain. | 2 / 3 |
Total | 8 / 12 Passed |
Validation
100%Checks the skill against the spec for correct structure and formatting. All validation checks must pass before discovery and implementation can be scored.
Validation — 11 / 11 Passed
Validation for skill structure
No warnings or errors.
- Repository
- pskoett/pskoett-ai-skills
- Commit
f6c5d7b
- 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.