docs(skills): remove dated prompt patterns from bundled skills - #367
Conversation
A prompt audit of the bundled skills turned up instructions that outlived their context. The publish skill told OpenClaw and QoderWork to swap their host id using a "supported list" that disappeared when the skills moved to per-agent templates, the find-skills skill repeated its keyword rule five times and carried circular tie-break coaching and a hardcoded "four options" that is wrong when only two exist, and _llm-client.md_ taught a hand-rolled JSON prompt, parse, and retry loop that `oo llm json` already provides. This removes the dead instructions, states each rule once, points structured output at `oo llm json`, aligns the description guidance in _oo-powered.md_ with _skill-authoring.md_, and adds the `bun run knip` step that _AGENTS.md_ was missing relative to _CLAUDE.md_. Signed-off-by: Kevin Cui <bh@bugs.cc>
|
Navigate logical layers of code changes, visualize relationships, and explore their blast radius. No actionable comments were generated in the recent review. 🎉 ℹ️ Recent review info⚙️ Run configurationConfiguration used: Organization UI Review profile: CHILL Plan: Essentials Run ID: 📒 Files selected for processing (8)
💤 Files with no reviewable changes (1)
Included review availability: 3 reviews are currently available. Your included PR review attempts over the past 7 days set your current allowance at 5 reviews per hour. Summary by CodeRabbit
WalkthroughSkill guidance now refers authors to the description trigger contract and directs runtime code that needs schema-valid JSON to use Priority: ⬇️ Low Merge Risk: ⚪ Minimal · up to The updated guidance and checks appear consistent, and declined installations still receive one short acknowledgement before the response stops. No actionable merge risk remains beyond normal checks. 🚥 Pre-merge checks | ✅ 4✅ Passed checks (4 passed)
✨ Finishing Touches✨ Simplify code
Comment |
A prompt audit of the bundled skills found instructions that outlived their context. The worst one tells OpenClaw and QoderWork to replace their host id with one "from the supported list", a list that vanished when the skills moved to per-agent templates, so the model is invited to rewrite an already correct
--agentvalue.oo-cli/contrib/skills/shared/oo-publish-skill/SKILL.md
Lines 86 to 90 in 72d33b0
The rest is the same kind of cleanup. oo-find-skills stated its keyword rule five times, carried a circular tie-break rule and a "four options" recap that is wrong when only two options exist, and llm-client taught a hand-written JSON prompt, parse, and retry loop that
oo llm jsonalready does with schema validation. oo-powered told authors to keep boundary conditions out of the description while skill-authoring asks for them, and AGENTS.md had drifted from CLAUDE.md by missingbun run knip. The embedded-assets test now asserts the rendered--agent qoderworkinstead of the removed sentence.