Content
92%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.
Excellent instructional content: fully actionable scripts, a sequenced workflow with explicit exit criteria and error-recovery guidance, and a clean one-level-deep reference structure. The only real cost is token efficiency — the receipts section duplicates the quotes reference and the constant guest-name citations could be thinned.
Suggestions
Remove or compress the "Receipts" section — the six quoted lines with timestamps duplicate references/quotes.md, which the "Go deeper" section already points to; keep at most one anchor quote in the body.
Thin the inline guest citations (e.g., collapse "(Zhang, Gupta, Jason Shah)"-style clusters to a single name) in plays A–G; the full attributions already live in references/frameworks.md, and "Grade existing work" already re-cites the relevant guests per criterion.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is dense, non-obvious expert guidance with almost no explanation of concepts Claude already knows, but there is trimmable fat: the "Receipts" section duplicates six quotes that live in references/quotes.md, and inline name citations ("(Fain)", "(Pfeffer, Terry)") appear on nearly every line. Not 3: the verbosity is minor padding around genuinely useful content; not 5: the receipts duplication and citation density are tokens that could be cut or moved to the reference file. | 4 / 5 |
Actionability | The skill gives copy-paste-ready material throughout: exact scripts ("Tell me what the board is pushing you on", "how strongly do you feel about this?"), a fill-in deliverable template with every cell specified, a 7-criterion grading rubric with point anchors, and named first moves. Not 4: guidance is executable end-to-end, covering the common cases (pitch, disagreement, overload, new-role trust). | 5 / 5 |
Workflow Clarity | The sequence is explicit (Diagnose with max-4 questions → Pick the play via routing table → Run plays → Deliverable), with validation-style checkpoints ("Done when every critical-path person has a stance, a failure fear, an evidence type and a named first move"), stop conditions ("Climb one rung at a time; stop when it resolves"), and error-recovery loops ("If the objection is fatal, rethink the pitch"). Not 4: exit criteria, error handling, and a completion checklist are all explicitly present. | 5 / 5 |
Progressive Disclosure | The body is a well-organized overview-plus-operating-instructions, with both bundle files (references/frameworks.md, references/quotes.md) verified to exist, clearly signaled one level deep in the "Go deeper" section with descriptions of what each contains. Not 4: navigation is easy, the split is appropriate (compressed plays inline, deep material in references), and there is no nesting or burying. | 5 / 5 |
Total | 19 / 20 Passed |