Content
88%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 lean, well-structured routing skill: an efficient protocol table, executable fingerprint commands for all four protocols, and an explicit identify-then-load workflow with a clean sibling pointer. The only notable gap is that the four sub-skills it routes to are named without any load path or command, which slightly weakens both actionability and navigation. It scores highly on conciseness and workflow clarity with no padding or concept over-explanation.
Suggestions
Show the load mechanism for the four sub-skills, e.g. a `load_skill("/skills/api/grpc/SKILL.md")`-style path in the table or one example line, so the routing action is fully executable end-to-end.
Add the concrete sub-skill paths (or a one-line note on where they live) so navigation to the core targets is as explicit as the sibling REST/GraphQL pointer.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The ~35-line body is lean and dense: a compact sub-skill table with wire signatures, four short fingerprint commands with minimal comments, and a two-line sibling pointer. It assumes Claude's competence, explains no background concepts, and every token earns its place — matching the top anchor rather than the 'minor trimming' level below. | 5 / 5 |
Actionability | The 'Quick fingerprint' section provides concrete, copy-paste-ready curl commands covering all four protocols (e.g. `curl -sk -I --http2 https://target/svc | grep -i 'application/grpc'`, the ?wsdl probe, the 101 upgrade check). The gap keeping it below fully-executable: the four sub-skills the routing table points to are named but no load path or command is given for them — the `load_skill(...)` pattern is only demonstrated for the sibling REST/GraphQL skill. | 4 / 5 |
Workflow Clarity | This is a simple single-purpose routing skill, and the single action is unambiguous: 'Identify the API protocol then load the matching sub-skill', backed by concrete identification commands per protocol. The routing table plus fingerprint makes the identify-then-load sequence explicit; the destructive/batch validation cap does not apply since no risky operations are involved. | 5 / 5 |
Progressive Disclosure | The skill is under 50 lines with well-organized sections (Sub-skills, Quick fingerprint, Sibling skills) and the one external reference (`load_skill("/skills/standard/exploit/web/SKILL.md")`) is a clearly signaled one-level path. However, the four sub-skills — the skill's core navigation targets — are referenced by name only, with no paths or load commands, a minor organization gap against the clear-navigation top anchor. | 4 / 5 |
Total | 18 / 20 Passed |