Content
77%Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
The body is highly actionable with a clearly sequenced, validated workflow for a fragile scraping task, but it pays for it in conciseness (repeated token/session/coupling explanations) and keeps everything inline rather than splitting deeper detail into signaled reference files. De-duplicating the repeated facts and moving the innerText-parsing detail into a reference would lift the weakest dimensions.
Suggestions
State the token source, session-sharing/concurrency rules, and published-track integration rationale each in exactly one place, then cross-reference rather than re-explaining.
Move the innerText parsing structure and detailed integration notes into a reference file (e.g. references/parsing.md) and signal it explicitly from the body, rather than inlining it in SKILL.md.
Explicitly point to scripts/fetch_engagement.py for the executable implementation so the SKILL.md reads as an overview rather than a monolith.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is dense and operational with no conceptual padding, but several points are repeated across sections — the token source is explained in 关键发现 #2, fetch 流程 step 4, and Notes; session-sharing/concurrency appears in 前置条件 #1, 关键发现 #3, and 约束; and the 4-step internal flow is restated in '与 published-track 集成'. It could be tightened by stating each fact once. | 2 / 3 |
Actionability | Provides concrete, copy-paste-ready CLI commands with flags and exit codes, exact URLs with the required token parameter, a 9-step fetch flow with real camoufox-cli invocations, a full output JSON example, and a complete executable script (scripts/fetch_engagement.py). | 3 / 3 |
Workflow Clarity | The fetch flow is clearly sequenced with an explicit validation checkpoint (wx-mp-hunter check exit 0/2 branching), consistent exit-code semantics, and recovery loops in Pitfalls (DOM change → probe, token expiry → re-login, rate limit → manual fallback) for this fragile browser-scraping + DB write operation. | 3 / 3 |
Progressive Disclosure | Sections are clearly organized and a bundle script exists, but the SKILL.md is a ~196-line monolith that inlines detail (innerText parsing structure, integration rationale, all pitfalls) rather than splitting it into well-signaled one-level-deep reference files; the script is not explicitly referenced for the parsing logic. | 2 / 3 |
Total | 10 / 12 Passed |