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.
An efficient, well-structured hub skill: concise routing tables, concrete tool and path guidance, and explicit verification and delegation rules. The referenced workflow and tool files carry the detailed steps, so validation loops and example invocations are not visible from this file alone.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The ~44-line body is lean and dense: routing tables, tool names, storage paths, and operational constraints with zero padding and no explanation of concepts Claude already knows. Every token earns its place, matching anchor 5. | 5 / 5 |
Actionability | Concrete, executable guidance throughout — a task-to-workflow routing table, exact tool names and commands ("douyin-hunter live", "info-record", wrapper "init-db"), and precise storage paths ("osbi/live/<platform>/<日期时间>-<room_id>/"). Not 5 because full command usage is deferred to external tool docs with no example invocations for the common cases. | 4 / 5 |
Workflow Clarity | The opening gives an explicit sequence ("先明确研究问题、对象、范围与交付形式,再取证、提取、交叉核实和分析") and the analysis section sets verification checkpoints (attach sources, record sample scope, explicitly flag unverifiable info). Falls short of 5 because detailed workflow steps and error-recovery loops are deferred to workflow files and no validate→fix→retry loop is stated in the body. | 4 / 5 |
Progressive Disclosure | Well-signaled, one-level-deep references (workflows/*.md links, tools/*/SKILL.md links) keep the body a clean routing overview. Minor gaps: the referenced bundle files are not present to verify, and "browser-guide" plus its nested "references/source-evidence.md" path is referenced by name only without a resolvable link. | 4 / 5 |
Total | 17 / 20 Passed |