Content
67%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, contract-driven skill body with a genuinely useful worked example and correctly externalized templates. Its main weaknesses are repetition of the ref/authorization rules across multiple sections and dense multi-clause steps that bury otherwise clear workflows.
Suggestions
State the opaque-ref / transient-render / distribution-authorization rules once (e.g., in the Skill Contract 'Writes' bullet) and reference that single statement from instructions 5-6 instead of restating it nearly verbatim three times.
Split instruction 6 into sub-steps (return inline → offer WARM save → transient render → distribution gate) so each authorization condition is a discrete, checkable step rather than one dense paragraph.
Add a brief final 'Done when' self-check step at the end of the workflow so the agent validates the finished report against the five completion criteria before returning it.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body never explains concepts Claude already knows and is mostly on-message, but the ref/authorization rules are restated nearly verbatim in three places (the **Writes** bullet, "Done when" item 5, and instruction 6: "exact WARM-save authorization", "never save the mapping", "requires a new exact authorization"), which is tighten-able padding. Not 'noticeably verbose with several unnecessary explanations' since all repetition is project-specific constraint, but clearly below 'efficient'. | 3 / 5 |
Actionability | As an instruction-only skill it gives concrete, executable guidance: a copy-paste Quick Start invocation, a full worked example with a rendered output table and metric-status prose, concrete save paths (`memory/influencer/report-generator/YYYY-MM-DD-<topic>.md`), and explicit NEEDS_INPUT fallback behavior. Minor gap: instructions 5-6 pack heavy conditional logic into dense prose rather than discrete steps, leaving the user to untangle authorization semantics. | 4 / 5 |
Workflow Clarity | The 7 numbered instructions form a clear sequence (determine parameters → pick template → apply guidance → recommendations → preserve Retro Card → return/save → offer next) with explicit validation checkpoints ("If the required computed artifact is absent, return a partial shell plus NEEDS_INPUT", card "absent, mismatched, or stale" checks). Falls short of a 5 because step 6 bundles ~six conditional rules in one step and there is no final output self-check against the 'Done when' list. | 4 / 5 |
Progressive Disclosure | Good structure: templates, visualization recs, and writing guidance are correctly split into the real one-level-deep reference `references/report-templates.md` (verified present and substantive), each reference is listed with a description in 'Reference Materials', and instructions link to it at the right moments. Below a 5 because ~40 lines of Skill Contract prose are inlined that could live in a reference, and many sibling-skill links (e.g. `../../../references/skill-contract.md`, `../../target/campaign-planner/SKILL.md`) are not resolvable within this bundle. | 4 / 5 |
Total | 15 / 20 Passed |