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 a dense, router-specific spec with concrete load/run commands, an explicit resolution gate, and a fallback, making the core sequence easy to follow. Its main weaknesses are triple restatement of the same scope limit, error paths deferred entirely to external modules, and a complete disconnect between the body's references and the shipped scripts/ bundle.
Suggestions
Remove the near-verbatim duplication of the scope limit — either drop the intro paragraph or trim RULES 36/37 to a single boundary statement — since PURPOSE and one RULE already carry the same content.
Connect the router to the shipped bundle: state how {cfs_cmd} maps to scripts/studio.py (or reference the relevant scripts/studio/commands), so SKILL.md navigation matches the actual bundle structure.
Add one inline checkpoint or fallback for the ambiguous-intent and no-matching-workflow cases instead of deferring them wholly to the external root-intent-routing module.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The scope limit ("resolves available cf-* workflows ... presents a launch menu") is stated near-verbatim three times — in the intro paragraph, in PURPOSE ("only resolves workflows, captures intent when needed, and presents a launch menu"), and in RULES ("ALWAYS limit cf/cf-studio to resolving available cf-* workflows, capturing unclear intent, matching workflows/companions, and presenting a launch menu") — and RULES 36/37 overlap on workflow-specific work being out of scope. Mostly efficient but could be tightened, matching the level-3 anchor; not level 4 because the repetition goes beyond minor. | 3 / 5 |
Actionability | The DO block gives concrete, executable-within-the-system guidance: full module paths to LOAD, a specific resolution source ("RESOLVE {cf-studio-path} from the in-context `@cf:root-agents` rule (fallback: its block in the project-root `AGENTS.md`)"), and named RUN steps. Not level 5 because execution semantics are deferred to externally loaded modules and the intro's "companion-aware" matching is never operationalized in DO. | 4 / 5 |
Workflow Clarity | The sequence is clear and gated: "REQUIRE {cf-studio-path} is resolved before any LOAD below" is an explicit checkpoint, the resolution step has a fallback, and steps end in a defined handoff (EMIT load report, CONTINUE IntentRouting). Not level 5 because the ambiguous-intent and no-match error paths have no inline checkpoint — they are only deferred to the external IntentRouting module. | 4 / 5 |
Progressive Disclosure | The six module LOADs are one level deep and clearly signaled, but they resolve through the runtime variable {cf-studio-path} to files outside this skill's bundle, while the actual shipped bundle (scripts/studio.py and 40+ modules under scripts/studio/, including commands like pdsl.py and validate.py) is never referenced from the body. Scored against the actual bundle structure, navigation from SKILL.md to real bundle content is missing, which fits the level-3 anchor ('could be better organized'); the clearly signaled references keep it above level 2. | 3 / 5 |
Total | 14 / 20 Passed |