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.
61
71%
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 articulates specific capabilities and includes an explicit 'Use when' trigger clause. Its main weakness is that the trigger terms lean toward specialized jargon ('gh-aw', 'structured learning candidates') rather than natural language users would employ when seeking this functionality. The description is distinctive and would not easily be confused with other skills.
Suggestions
Expand trigger terms to include more natural user language variations such as 'CI/CD', 'continuous integration', 'build failures', 'automated pipeline checks', or 'GitHub Actions failures'.
| Dimension | Reasoning | Score |
|---|---|---|
Specificity | Lists multiple specific 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 distinct, concrete capabilities. | 3 / 3 |
Completeness | Clearly answers both 'what' (captures failure patterns, emits learning candidates, proposes prevention rules) and 'when' with an explicit 'Use when:' clause specifying 'automated learning capture in CI/headless pipelines.' | 3 / 3 |
Trigger Term Quality | Includes some relevant terms like 'CI', 'pull request checks', 'headless pipelines', 'gh-aw', and 'GitHub Agentic Workflows', but misses common user variations like 'CI/CD', 'continuous integration', 'automated checks', 'build failures', or 'pipeline errors'. The term 'gh-aw' is niche jargon that most users wouldn't naturally say. | 2 / 3 |
Distinctiveness Conflict Risk | Highly distinctive with a clear niche: CI-only, non-interactive, self-improvement workflow tied to GitHub Agentic Workflows. The combination of 'CI-only', 'headless', 'gh-aw', and 'failure patterns' makes it very unlikely to conflict with other skills. | 3 / 3 |
Total | 11 / 12 Passed |
Implementation
57%Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
The skill provides good structural organization and a useful output schema, but falls short on actionability—the core CI workflow (inspecting checks, deduplicating patterns, emitting candidates) is described conceptually rather than with concrete, executable steps. The recurrence rules and promotion targets are well-defined, but Claude would need more specific guidance on how to actually implement the inspection and deduplication logic.
Suggestions
Add concrete, executable code or command sequences for the core workflow: how to inspect PR check results (e.g., specific gh CLI commands), how to parse and deduplicate learning candidates, and how to emit the output schema.
Include a validation/feedback loop for the main CI process—e.g., what to do when candidate ingestion fails, how to handle malformed input from simplify-and-harden-ci, or how to verify the emitted output is valid.
Remove or condense the 'Context Limitation' and 'CI Contract' sections into a brief constraints list—Claude doesn't need the rationale explained at length.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Generally efficient but includes some unnecessary sections like the 'Context Limitation' explanation and the 'CI Contract' which describe constraints Claude could infer. The install section at the top also adds tokens for something that isn't core to the skill's instructional content. | 2 / 3 |
Actionability | Provides a concrete output schema and some executable commands (gh aw compile, gh aw run), but the core workflow—how to actually inspect PR checks, ingest candidates, deduplicate patterns, and emit output—is described abstractly rather than with executable code or concrete step-by-step commands. The skill tells Claude what to do conceptually but not how to do it mechanically. | 2 / 3 |
Workflow Clarity | The authoring workflow section has a clear 4-step sequence with validation, but the main CI workflow (inspect → ingest → deduplicate → emit) lacks explicit sequencing and validation checkpoints. For a skill involving batch/automated operations in CI, the absence of error recovery or feedback loops in the core process caps this at 2. | 2 / 3 |
Progressive Disclosure | Content is well-structured with clear sections, references example templates in 'references/workflow-example.md' without deep nesting, and appropriately points to related skills (simplify-and-harden-ci, self-improvement) for extended functionality. The organization is clean and navigable. | 3 / 3 |
Total | 9 / 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
fe0da1c
- 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.