- Home
- Skills
- Omidzamani
- Dspy Skills
- Skill Perfection
skill-perfection_skill
- Python
26
GitHub Stars
1
Bundled Files
2 months ago
Catalog Refreshed
4 months ago
First Indexed
Readme & install
Copy the install command, review bundled files from the catalogue, and read any extended description pulled from the listing source.
Installation
Preview and clipboard use veilstrat where the catalogue uses aiagentskills.
npx veilstrat add skill omidzamani/dspy-skills --skill skill-perfection- SKILL.md6.4 KB
Overview
This skill provides a repeatable QA methodology for auditing and fixing plugin skill files in a single pass. It combines an optional Python preflight check with an LLM-driven audit+fix workflow, spot verification, and a concise verification report. The goal is to eliminate redundant iterations and produce a clear PERFECTION_REPORT.md for each skill.
How this skill works
Start with an optional preflight script for Python-heavy skills to catch obvious syntax issues, then run an LLM-guided Audit+Fix pass that inspects the file section-by-section. When issues are found, fix them in-place immediately, log the change, and continue. After the single-pass audit, spot-check a few modified items and generate a short verification report summarizing fixes, sources, and status.
When to use it
- You need a fast, authoritative audit and in-place fixes for a plugin skill file.
- The skill contains Python code blocks and you want an optional preflight syntax check.
- You want a single-pass workflow to minimize rework and web searches.
- You must produce a concise verification report for reviewers or CI.
- You want to align examples and API usage with official documentation.
Best practices
- Use the preflight helper only for Python-heavy skills; treat its output as advisory.
- Batch related documentation lookups to reduce web calls and duplicate research.
- When you find an issue: stop, fix immediately, record {location, old, new, reason, source_url}, then continue.
- Prioritize official docs, then official repos, package registries, and changelogs when researching.
- Spot-check 3–5 changes after the pass instead of re-auditing the whole file.
Example use cases
- Audit a Python-based Claude skill with multiple code examples and update incorrect imports.
- Fix outdated API calls in a JavaScript plugin and align signatures with official docs.
- Verify and correct broken or outdated URLs referenced in the skill file.
- Produce a PERFECTION_REPORT.md summarizing all fixes and sources for maintainers.
- Use the workflow in CI to quickly triage and fix simple syntax or doc-errors before release.
FAQ
Ignore preflight if it produces confusing errors, more than ~10 irrelevant errors, tracebacks from the script, or if it hangs; proceed to the LLM audit.
How many web searches are allowed?
Target fewer than 20 searches per skill by batching related lookups; prioritize official documentation and authoritative sources.