Content
63%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 an efficient, well-organized set of during-session heuristics with genuinely useful example phrasings and a disciplined ADR gate. Its main defect is that both of its format references dangle — CONTEXT-FORMAT.md and ADR-FORMAT.md do not exist — which degrades actionability and leaves the progressive-disclosure structure unverifiable.
Suggestions
Create CONTEXT-FORMAT.md and ADR-FORMAT.md (or inline their formats into SKILL.md) so the instructions 'Use the format in [CONTEXT-FORMAT.md]' and 'Use the format in [ADR-FORMAT.md]' resolve to real content — as written, a session cannot execute the record/update steps.
Add a brief explicit ordering of the session moves (challenge → sharpen → cross-reference with code → record resolved term → evaluate ADR gate) so the flow reads as a sequence rather than parallel heuristics.
Trim the two ASCII directory trees to a single compact example (e.g. show the multi-context layout only, and describe the single-context case in one line) to cut ~10 lines of visual bulk.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is lean and assumes competence — terse sections, illustrative one-line dialogue quotes, a crisp three-condition ADR test — with every token earning its place. Falls short of 5 because the two full ASCII directory trees (~25 lines) and the meta-parenthetical distinguishing this skill from passive CONTEXT.md reading could be compressed without losing guidance. | 4 / 5 |
Actionability | The conversational moves are concrete ("Your glossary defines 'cancellation' as X, but you seem to mean Y — which is it?"), but the operative formatting guidance is delegated to "Use the format in [CONTEXT-FORMAT.md](./CONTEXT-FORMAT.md)" and "[ADR-FORMAT.md](./ADR-FORMAT.md)" — files that do not exist in the bundle. That leaves key execution details missing, matching the anchor for incomplete concrete guidance rather than the mostly-executable level above. | 3 / 5 |
Workflow Clarity | Each move has an explicit trigger condition ("When the user uses a term that conflicts...", "When a term is resolved...") plus lazy-creation rules for CONTEXT.md and docs/adr/, giving a discernible detect → challenge → resolve → record flow. Not 5 because the moves are presented as parallel heuristics with no explicit ordering or checkpoint tying them together, and the record step is undercut by the missing format files. | 4 / 5 |
Progressive Disclosure | References are well-signaled one-level-deep links and the body is sensibly sectioned, but both referenced files (CONTEXT-FORMAT.md, ADR-FORMAT.md) are absent from the bundle — there is no references/, scripts/, or assets/ directory at all. Navigation to the detailed material fails outright, so the disclosure structure is present on paper but broken in fact, fitting the mid anchor rather than the well-placed level above. | 3 / 5 |
Total | 14 / 20 Passed |