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, action-oriented skill body with clear escalation workflow and validation discipline, organized into clean sections without external file dependencies. Minor conciseness trims and a concrete example call would push it to the top tier.
Suggestions
Tighten the vendor-name lists (e.g. 'Cloudflare, Akamai, DataDome') into one consolidated mention to save tokens without losing signal.
Add one concrete web_fetch example call (e.g. web_fetch(url, selector='#content')) so the recommended selector/device knobs are shown in use.
Consider moving the platform fast-path table into a reference file if the table grows, to keep the overview lean.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Mostly lean and assumes Claude's competence, with tight rule framing (R2, R6); a few explanatory passages about WAF vendors and escalation phrasing could be trimmed slightly. | 4 / 5 |
Actionability | Provides concrete guidance (knobs: selector/device, fast-path table, specific retry recipe with device='mobile'), but no executable code/command examples — acceptable for this instruction/tool skill, with only minor gaps. | 4 / 5 |
Workflow Clarity | Clear sequenced flow (web_search -> web_fetch), explicit escalation ladder with phases, and a robust validation/retry feedback loop (terminal vs non-terminal stop_reason, retry with mobile+selector) — strong checkpoints. | 5 / 5 |
Progressive Disclosure | Well-organized single-file skill (~90 lines) with clear section headers and no nested references; appropriately self-contained for a skill that needs no external bundle files. | 5 / 5 |
Total | 18 / 20 Passed |