Content
81%Weight 40%Scale 1-5Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
This is a strong, highly actionable skill: executable Groovy and exact script invocations throughout, plus a genuine validate-by-severity-and-re-validate feedback loop. The main weakness is token efficiency — basic syntax examples Claude already knows and triplicated guidance (fingerprint, failFast) pad the file — and the syntax reference could be better split into the existing bundle files.
Suggestions
State the fingerprint and failFast requirements once each in their canonical sections (Post and Parallel & Matrix) and remove the duplicate mentions in the Declarative process list and Post commentary.
Replace the trailing '## References' section's minimal-pipeline, catchError, and deploy examples plus the options table with pointers to references/best_practices.md and assets/templates/, since that material duplicates earlier sections.
Move the generic 'Minimal Declarative Pipeline' and basic scripted node/try-catch examples into assets/templates/ (they are boilerplate Jenkins knowledge), keeping SKILL.md focused on the non-obvious guidance like durabilityHint, input-outside-steps, and NonCPS constraints.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is mostly efficient reference material, but includes content Claude already knows ("pipeline { agent any ... stage('Build') { steps { sh 'make' } }"" minimal example, generic node/try-catch scripted syntax) and repeats critical rules — "Always include `fingerprint: true` when using `archiveArtifacts`" appears three times, and the trailing "## References" section re-covers catchError/timeout/buildDiscarder already detailed earlier. This fits the score-3 anchor ('mostly efficient but includes some unnecessary explanation or could be tightened'); it is not score 2 because there is no large-scale padding or tutorial-style explanation of basics. | 3 / 5 |
Actionability | Guidance is fully executable and copy-paste ready: complete Groovy snippets for every feature ("parallelsAlwaysFailFast()", matrix axes/excludes, kubernetes yaml pod) and exact commands like "python3 scripts/generate_declarative.py --output Jenkinsfile --stages build,test,deploy --agent docker" and "bash scripts/validate_jenkinsfile.sh --syntax-only Jenkinsfile". Common cases are covered with specific examples, matching the score-5 anchor and exceeding score 4's 'minor gaps'. | 5 / 5 |
Workflow Clarity | Multi-step processes are clearly sequenced with explicit validation checkpoints and a feedback loop: "1. Generate Jenkinsfile 2. Invoke devops-skills:jenkinsfile-validator 3. Handle validation results by severity... 4. Re-validate after fixes 5. Present validated Jenkinsfile". This matches the score-5 anchor (clear sequence, explicit validation, error-recovery loop); score 4 would lack the re-validate feedback step, which is present. | 5 / 5 |
Progressive Disclosure | The bundle is well-structured and all referenced paths are real one-level-deep files (references/best_practices.md and common_plugins.md exist with substantive content, as do assets/templates/declarative|scripted/basic.Jenkinsfile and the three generator scripts), and the body signals them with purpose notes ("references/best_practices.md - Performance, security, reliability patterns"). It falls short of score 5 because a sizable inline syntax reference (Declarative/Scripted/Docker-K8s sections) plus a trailing generic example section could largely live in a reference file, leaving SKILL.md as a leaner overview with clearly signaled pointers. | 4 / 5 |
Total | 17 / 20 Passed |