Content
67%Weight 40%Scale 1-5Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
The body is a well-structured reference-adaptation guide: a concrete file inventory, six specific porting gaps with exact interface names, an executable shared-CLI command, and explicit pre-test safety checklists. It is held back by repeatedly restated 'do not run this' boundary disclaimers that could be consolidated, and by pointing to an authoritative README.md that is not present in the provided skill bundle.
Suggestions
Consolidate the repeated reference-only/do-not-run caveats (intro, end of the porting gaps, shared-engine section) into a single boundary statement near the top to reclaim tokens.
Ship the README.md the body calls authoritative (or inline its essential adaptation checklist) so the skill's primary navigation path resolves within the bundle.
Add a short validate→fix→retry loop for the porting workflow (e.g. run the mock-backend end-to-end tests after each gap is closed, and only proceed when they pass) to complete the feedback loop.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The reference-only/do-not-run caveat is repeated at least four times ("not a supported, plug-and-play OpenClaw integration", "Do not run or schedule the files unchanged", "treat this directory only as source material", "Use the shared documentation for those features, not this reference SKILL"), and the data-boundary section partially restates porting-gap concerns — clearly tighten-able. It is not 2 because no space is spent explaining concepts Claude already knows; all content is project-specific, and not 4 because the repeated boundary disclaimers exceed minor trimming. | 3 / 5 |
Actionability | Concrete, specific guidance throughout: an executable command ("python -m skillopt_sleep" with "--backend mock"), a file inventory with roles, exact interface names ("get_backend" call contract, "skillopt_sleep.staging.adopt", "proposed_SKILL.md" / "proposed_CLAUDE.md", "manifest.json"), an enumerated backend list (mock, claude, codex, copilot, handoff, azure_openai), and six numbered porting tasks with technical detail. Not 5 because no copy-paste-ready code for the porting work and the actual adaptation checklist is delegated to README.md; not 3 because this is far more concrete than 'some guidance with missing key details'. | 4 / 5 |
Workflow Clarity | "Before treating this as an integration, a maintainer must at least:" introduces a clearly sequenced six-step porting process ending in a validation step ("Add isolated end-to-end tests for dry-run, accepted/rejected gates, staging, adoption and backup, credential failure, and scheduled execution"), plus an explicit pre-test checklist in the data-boundary section. Not 5 because no explicit validate→fix→retry feedback loop is defined and the operational workflow is delegated to README.md; not 3 because validation checkpoints are present and explicit rather than implicit, and the destructive/batch cap does not apply to a porting guide. | 4 / 5 |
Progressive Disclosure | Well-organized sections (What is included / Known porting gaps / Shared-engine features / Data and credential boundary / Further information), a clear file inventory, and well-signaled one-level-deep links with descriptions. Not 5 because the body directs readers to "Start with the directory's README.md, which is the authoritative status and adaptation guide", yet no README.md or other bundle files (no references/, scripts/, assets/) ship alongside this SKILL.md, so navigation does not fully resolve; not 3 because references are clearly signaled, not buried, and the ~100-line body is appropriately split into scannable sections. | 4 / 5 |
Total | 15 / 20 Passed |