Content
42%Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
The skill is highly actionable with concrete CLI commands and clear branching logic for different violation types, but it is severely undermined by extreme verbosity and repetition. The same notification comment is duplicated four times, and sanitization principles are restated across multiple sections. The lack of any progressive disclosure or content extraction into supporting files makes this a monolithic document that wastes significant context window space.
Suggestions
Extract the notification comment text into a single constant/template referenced once, rather than duplicating it four times with implementation blocks.
Consolidate the three violation sections (title, body, comments) into a single parameterized workflow since they follow the same pattern: sanitize → update via gh CLI → post notice → report.
Remove the enumerated list of violation types (hate speech, slurs, etc.) — Claude already knows these, and the skill already instructs reading CODE_OF_CONDUCT.md which would define them.
Extract report format templates and the notification message into a separate TEMPLATES.md file to reduce the main skill's size and improve progressive disclosure.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Extremely verbose and repetitive. The same notification comment text is duplicated four times verbatim (for title, body, and comments sections, plus implementation blocks). The same sanitization principles are restated multiple times. The skill could be reduced to roughly 1/3 its size without losing any information. It also explains concepts Claude already understands (what hate speech is, what trolling is, etc.). | 1 / 3 |
Actionability | Provides fully executable gh CLI commands for every action (editing titles, bodies, comments, posting replies). The commands use proper API endpoints and flags. Examples of title replacement logic are concrete and illustrative. The report formats are copy-paste ready. | 3 / 3 |
Workflow Clarity | The process steps are listed and sequenced (fetch comments → evaluate → take action → report), but there are no explicit validation checkpoints. For a moderation task that involves destructive edits (replacing content permanently), there's no verify-before-commit step or rollback guidance. The workflow branches clearly by violation type but lacks feedback loops for error cases (e.g., what if the gh command fails). | 2 / 3 |
Progressive Disclosure | This is a monolithic wall of text at ~200+ lines with no references to external files and massive inline repetition. The notification text, sanitization principles, and reporting formats could be extracted into templates or referenced files. The content is poorly organized with redundant sections that inflate the document significantly. | 1 / 3 |
Total | 7 / 12 Passed |