Content
77%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 a well-structured, highly actionable guide with an explicit workflow, validation checklist, feedback-style troubleshooting, and clean one-level-deep references to real bundle files. Its main weakness is token efficiency: core ideas (claim framing, refrain handling) are restated across three or four sections and could be consolidated.
Suggestions
Consolidate the claim-framing guidance: Mindset, Prerequisites, Workflow step 1, and the first Core Mechanics bullet all restate 'an obvious truth the audience was too polite to say' — state it once and cross-reference.
Merge refrain guidance into one place: it currently appears in Workflow step 4, Core Mechanics, Example 3, and Troubleshooting; a single treatment with the example would cut significant tokens.
Replace the five isolated technique snippets with one complete short bit annotated by technique, plus 1-2 fragments — this both tightens the Examples section and closes the actionability gap of never showing a full assembled routine.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is mostly efficient domain guidance Claude would not produce unprompted (contrastive framing, escalation ordering, objector handling), but the same points recur across sections: the 'obvious truth the audience was too polite to say' framing appears in Mindset ("the audience already half-believed the claim"), Workflow step 1, and Core Mechanics' opening bullet; refrain guidance is repeated in Workflow step 4, Core Mechanics, Examples, and Troubleshooting ("check whether they're actually escalating or just restating"). This fits anchor 3 ('mostly efficient but could be tightened') better than anchor 4, since the repetition is more than minor. | 3 / 5 |
Actionability | Concrete, executable guidance throughout: an 8-step Workflow ("Draft 3-5 escalating examples... ordered so each one is more pointed than the last"), three BAD/GOOD example pairs with SYMPTOM and CONSEQUENCE, and five tagged example snippets each with a '*What it demonstrates*' note. This is mostly anchor 4-5 territory for an instruction-only skill; it falls short of anchor 5 because no complete end-to-end example bit is shown — all demonstrations are fragments in isolation, leaving the assembly (especially pacing across a full routine) to be inferred. | 4 / 5 |
Workflow Clarity | The Workflow section gives a clearly numbered 8-step sequence with an explicit validation step — "Check the draft. Run through the checklist in ## Troubleshooting" — and the Troubleshooting section closes the loop with symptom→fix guidance ("If the closing line's edge is missing, check for a hidden qualifier and cut it") plus a four-item delivery checklist. Prerequisites even adds an early user check ("Read back the claim in one sentence before drafting examples, so the user can redirect early"), matching anchor 5's explicit validation and feedback loops. | 5 / 5 |
Progressive Disclosure | The body is a well-organized overview and both referenced files exist and are one level deep, clearly signaled in a When-to-Use table ([Refrain and Escalation Techniques], [Subject Selection and Framing]) plus targeted inline links from Troubleshooting ("see [Subject Selection and Framing](references/subject-selection-and-framing.md)"). The deep-dive material (refrain mechanics, subject framing) is appropriately split out rather than inlined, matching anchor 5; the only blemish is that the refrain reference is linked twice in the References section (bullet plus table row), which is cosmetic. | 5 / 5 |
Total | 17 / 20 Passed |