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.
The body is an exemplary lean operations index: dense zero-padding prose with concrete tool names, paths, artifacts, role assignments, and explicit verification gates covering reads, writes, publishing, and review. Structure routes every task to a one-level-deep workflow file; the only notable weakness is that the referenced bundle files were not present to verify, and per-workflow detail lives entirely in those files.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The ~30-line body is purely operational — tool routing ("tiktok-publish 发布,tiktok-engagement 本人取数回填"), storage paths ("tiktok/ref/、tiktok/outputs/、tiktok/calibration/"), constraints ("复盘只用实际指标,缺项不补零") — with zero concept explanations or padding, matching the 'lean and efficient; every token earns its place' anchor. Not 4 because there is no over-explanation to trim; every sentence carries non-obvious operational state. | 5 / 5 |
Actionability | Concrete, executable guidance throughout — named tools invoked "通过 wrapper 调用", exact artifacts ("dna-meta.json", "published-track"), ordered operations ("视频样本先 hunter 下载,再运行 viral-chaser"), role splits (main 负责选题/Brief vs CP 负责旁白与制作) — matching 'mostly executable guidance; concrete commands with minor gaps'. Not 5 because no literal commands or worked examples appear and exact tool invocation is deferred to external tool docs; not 3 because the specifics given are real and directly executable, not pseudocode. | 4 / 5 |
Workflow Clarity | The task table routes each work type to a workflow file and checkpoints are mostly explicit — "读取与写入均需真实账号验收", "完成后检查成品", "所有发布返回真实作品 ID/URL 后记录 published-track", "写操作需要同次 ticket-guard 材料" — matching 'clear sequence with most checkpoints present; minor validation gaps'. Not 5 because per-workflow step sequences and error-recovery loops are delegated to the referenced files rather than stated as step-gated lists; not 3 because validation requirements are explicitly written, not missing or only implied. | 4 / 5 |
Progressive Disclosure | The body is a clean overview: a task→workflow index table (Account Setup, Style DNA, Content Production, Native UI Cards, Account Benchmark, Editing, Review) with one-level-deep references, and tool detail delegated to in-package docs ("读取本包 tools 下对应说明"), matching 'good structure; references mostly clear; minor organization gaps'. Not 5 because the referenced workflows/*.md files and tools docs could not be verified — the provided bundle contains only SKILL.md — leaving the primary navigation targets unconfirmed. | 4 / 5 |
Total | 17 / 20 Passed |