Content
78%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 instruction-only skill: lean body, concrete BAD/GOOD examples that teach the technique, a sequenced workflow with a final anti-pattern check, and exemplary conditional references. The main weakness is the fabricated 'make check-draft' command — non-executable content that weakens both actionability and the escalation-validation step — plus the lack of any error-recovery guidance after the final check fails.
Suggestions
Delete the illustrative bash block and its disclaimer ('make check-draft FILE=draft.md' / 'That command is illustrative only — this skill ships no such script'); replace step 6's escalation test with a concrete inline check, e.g., 'For each step in the chain, ask: could a real company do this under an existing terms-of-service clause? If not, rewrite the step'.
Add one-line recovery guidance after the step-9 final check (e.g., 'If a beat matches an anti-pattern, rewrite that beat against the paired GOOD example and re-check — do not restart the draft'), turning the final check into a genuine feedback loop.
Trim the opening paragraph, which restates the frontmatter description ('a cynical column, monologue, short story... one sustained, escalating metaphor'), to a single sentence of unique framing.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is efficient and assumes Claude's competence — technique guidance is delivered through compact BAD/GOOD example pairs rather than explanations of what satire is. Not 5: the illustrative-only bash block ('make check-draft FILE=draft.md' followed by 'That command is illustrative only — this skill ships no such script') spends tokens on a command that cannot be run, and the opening paragraph partially restates the frontmatter description. | 4 / 5 |
Actionability | Guidance is mostly executable for a writing skill: concrete mechanics ('PREFER a single scaling metaphor system... AVOID a second unrelated joke'), worked text examples ('The app doesn't send notifications, it triages you...'), and specific BAD/GOOD drafts for each anti-pattern. Not 5: the only 'command' in the skill is explicitly non-existent, and the workflow steps ('Add a jargon-mockery beat', 'Return to the centre near the end, worse') are directives without a worked mini-example for the story branch's escalation. | 4 / 5 |
Workflow Clarity | A clear 9-step numbered sequence with explicit branch selection (step 2: rant vs story branch) and a final validation checkpoint (step 9: 'Run a final check against the anti-patterns below to verify the draft before calling it done'). Not 5: validation is a one-shot final check rather than a feedback loop — there is no guidance on what to do when a check fails (e.g., rewrite the offending beat vs restart), and the escalation-logic test points to a command that does not exist. | 4 / 5 |
Progressive Disclosure | Clear overview in SKILL.md with well-signaled, one-level-deep references: the References table maps each topic to a real file (both references/extended-metaphor-and-voice.md and references/dystopian-escalation-structure.md exist) with precise conditional load guidance ('Load only when writing a rant/column/monologue and the metaphor is drifting...'). Matches the anchor-5 structure exactly. | 5 / 5 |
Total | 17 / 20 Passed |