Content
82%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.
A well-structured, token-efficient routing/index skill body: a clear numbered selection procedure, disciplined conditional-loading rules, and a sensible safety section. Its weaknesses are the absence of any worked routing example or fallback behavior for unhandled requests, and a reference hierarchy that nests two-plus levels deep rather than staying one level from the entry point.
Suggestions
Add one short worked routing example (request → selected workflow) so a correct selection is unambiguous and anchor-5 actionability is reachable.
Specify fallback behavior when the workflow index doesn't cover a request or a referenced file is missing, providing an explicit error-recovery loop.
Flatten the reference hierarchy where possible — surface one-line pointers to key workflows directly in SKILL.md so navigation from the entry point stays one level deep.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The ~30-line body is lean and assumes Claude's competence: every sentence instructs ("Read `workflows/index/SKILL.md` completely", "Do not load unrelated sales workflows merely because they are present") with no padding, no concept explanations Claude already knows, and no wasted tokens. Matches anchor 5 (every token earns its place). | 5 / 5 |
Actionability | As an instruction-only routing skill the absence of code is not penalized; the guidance is concrete (exact files to read, read-before-act ordering, conditions for loading provider skills, 'use the workflow's evidence and fallback rules'). However, there are no examples of what a correct routing decision looks like in practice — a minor gap that fits anchor 4 (mostly executable guidance) rather than anchor 5's fully concrete common-case coverage. | 4 / 5 |
Workflow Clarity | The numbered 1–5 'Route the request' sequence is clear with explicit load conditions, and the safety section adds confirmation checkpoints before external actions ("Require appropriate confirmation before sending messages, updating CRM records"). Not anchor 5 because there is no error-recovery/feedback loop (e.g., what to do when the index doesn't cover a request or a named workflow is missing); routing is not destructive so the anchor-3 cap does not apply. | 4 / 5 |
Progressive Disclosure | The overview role is clear and every reference is explicitly signaled with conditions ("Read `workflows/index/SKILL.md` completely", "read them only when their stated conditions apply"). However, the reference chain is multi-hop — SKILL.md → workflows/index/SKILL.md → per-workflow SKILL.md → conditionally related skills — i.e., 2+ levels of nesting the rubric penalizes, and no bundle files were provided alongside this SKILL.md to verify the referenced paths exist. Good structure with minor organization gaps fits anchor 4; not 3 because signaling and load conditions are genuinely well done. | 4 / 5 |
Total | 17 / 20 Passed |