Optimize the utilities skill family

[potiuk]

Part of #1342 — apply the setup-family optimization recipe to the utilities family.

5 skills, 17,809 body tokens in total; 3 over the 200-token always-on budget, 0 over the 5,000-token body budget.

skill	body tokens (budget 5,000)	always-on tokens, est. (budget 200)	lines (cap 500)
report-framework-issue	4,627	~287 ⚠	404
skill-reconciler	4,438	~250 ⚠	438
list-skills	2,288	~265 ⚠	216
optimize-skill	3,998	~180	358
write-skill	2,458	~111	228

Body tokens are from docs/mode-economics.md; always-on figures are a chars ÷ 4 estimate of description + when_to_use. ⚠ marks a budget exceeded. Line counts include the generated pre-flight block.

Order of work (per the umbrella):

1. Trim the frontmatter of every skill marked ⚠ in the always-on column.
2. Split any over-cap body, moving sub-action-specific rules into sibling files byte-for-byte.
3. Wording pass on the rest, with headings, golden-rule headlines, code blocks and the pre-flight block kept byte-identical.
4. Run each touched skill's eval suite before and after, outside the sandbox.

Record any inconsistencies found but not fixed (eval-coupled wording, stale headings) in the PR description.

[anbv29]

Hi! I’d like to understand the current utilities skill family and propose one narrowly scoped optimization with a benchmark or regression check. Is there a good entry point for a first contribution?

[potiuk]

yeah, absolutely. Just use the optimization skill to help you there. The skill is basically leading you step-by-step and we have unit tests and evals (and the skill will run them) to make sure it all works as before.

[potiuk]

You. Can limit it to one skill initially.

[potiuk]

(BTW. For example it would be really fun to optimize the optimize skill with the optimize skill)

[anbv29]

Thanks! I’ll start with the optimize-skill utility itself and keep this first pass limited to that one skill. I’ll run the validator and evals before making changes, use the optimization workflow to identify a small behavior-preserving improvement, and open a focused PR with the before/after results. If there’s a particular optimization pass you’d like prioritized, please let me know.

[anbv29]

Post this on issue #1350 after opening the PR:

Thanks for the guidance. I’ve completed the optimization of the optimize-skill workflow, reduced its token cost from 3,995 to 3,024 tokens, and added eval coverage for all seven optimization passes. I’ve opened a PR for review: https://github.com/apache/magpie/pull/1379. Feedback is welcome!