luigidemasi commented on issue #41: URL: https://github.com/apache/camel-kit/issues/41#issuecomment-6336824231
> Comment by [luigidemasi](https://github.com/luigidemasi) on 2026-09-03 ([original](https://github.com/luigidemasi/camel-kit/issues/73#issuecomment-5524262147)). --- Closing after re-checking the premise against current `main` (`8d4206b0`): - The loading-overhead premise no longer holds. Guides are loaded selectively — `skills/camel-design/SKILL.md` carries a per-guide "When to Load" manifest — and subagent personas reference exact guide files rather than loading whole skills: `agents/code-quality-reviewer.md` lists the five `camel-validate/guides/*.md` paths and `agents/test-engineer.md` the four `camel-test/guides/*.md` paths. `camel-execute` resolves guide selectors only to individual installed guide files. The specific case in the description — a security reviewer forced to load the full `camel-validate` skill — no longer exists. - The acceptance criterion of three extracted checklists is not supported by the current tree. Testing knowledge is owned by `camel-test` (~1,250 lines across five guides, referenced by path from other skills); the one genuinely duplicated block was deduplicated in PR luigidemasi/camel-kit#204. Performance snippets are restated across guides, but each copy loads in a different pipeline phase, so extraction would not reduce context, and the two observed inconsistencies were one-line fixes (also PR luigidemasi/camel-kit#204). - Security is the one domain where duplication has a measurable maintenance cost (the same core rules in 6–9 artifacts, with observed drift between copies). That narrower extraction is tracked in luigidemasi/camel-kit#205. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
