haorantang97/LabKit / skills/skill-skill/modules/polish-rule-content
polish-rule-content
VerifiedRevise an existing AI skill or rule for clearer discovery, useful instructions, appropriate scope, and reliable execution. Use for a requested content improvement or a demonstrated behavioral problem; a review-only request produces findings without rewriting.
Revise existing skill content
This is a shared module of skill-skill. Read authoring guidance for the core criteria and official validation tools. Work on the user's selected version and preserve its functioning behavior and license.
Identify the requested change. Compare the current instructions with the intended behavior and concrete failure evidence. For review or plan requests, report findings only. For an authorized edit, make the narrowest useful revision without introducing a redundant confirmation step.
Check discovery separately from execution. The description should make the capability and likely trigger clear; add exclusions only for plausible misrouting. Do not force five segments, a word-count target, negative examples or a fixed vocabulary when they add no useful distinction.
Keep the entrypoint focused on shared purpose, essential constraints and useful references. Put substantial conditional detail in supporting files when it reduces unnecessary context. Do not wait for an arbitrary line threshold, split a simple file into a routing system, or add sections merely to satisfy a template.
Use numbers and deterministic steps when the workflow has a real measurable invariant. Do not invent line limits, minimum test counts, performance targets or version claims just to make a rule look specific. Detailed gates should have a concrete reason; distinguish user conventions from host requirements and optional recommendations.
Write direct, natural instructions. Preserve the user's voice where relevant. Remove generic encouragement, repeated context, unsupported authority claims and decorative explanation. Do not globally ban ordinary words or punctuation when they express the instruction well.
Inspect callers before renaming or removing supporting files. Preserve metadata and invocation policy unless the requested change requires otherwise. Keep permissions, external actions, source boundaries and plan-only scope explicit where they materially affect the workflow.
Validate with the current official validator when available, then check changed references and behavior relevant to the edit. Use realistic input and observable outcomes for substantive behavior changes; a matching heading or phrase alone is not behavioral validation. Report what was actually verified and any remaining client test.
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis. | Stars: 468k | Updated | Claude CodeCodex | Verified | Install |
| diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness. | Stars: 321k | Updated | Claude CodeCodex | Verified | Install |
| dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependencies | Stars: 321k | Updated | Claude CodeCodex | Verified | Install |
| executing-plansobra/superpowersUse when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is available | Stars: 321k | Updated | Claude CodeCodex | Verified | Install |