Content
73%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-engineered instruction skill: the workflow is unambiguous, checkpointed, and safe by construction (identity hygiene, authorization gates, Unknown-blocking), with real template support in the bundle. Its weaknesses are token efficiency — core invariants are restated four to five times each — and navigation redundancy, with three overlapping sibling-link lists and several references that do not resolve inside the bundle.
Suggestions
State each core invariant (opaque creator_ref handling, Suitability/commercial separation, 'never enters the SQS') exactly once in the Skill Contract and reference it from the numbered steps instead of restating it in the intro, steps 1-8, and the example — this would cut substantial length without losing clarity.
Collapse 'Reference Materials', 'Next Best Skill', and 'Related Skills' into a single routing section: all three list the same sibling skills (influencer-discovery, competitor-tracker, audience-mapper, outreach-manager) with slightly different framing.
Inline the minimum scoring substance needed to run steps 3-6 (or bundle star-benchmark.md / skill-contract.md at resolvable paths), since the body's key definitions — cohort benchmarks for STAR-S2 and the ranking 'declared rule' — currently live in files that do not exist in this skill's bundle.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is dense and free of general-knowledge padding, but the same rules are restated many times: the Suitability-vs-commercial separation appears in the intro, contract, steps 3/6/7, and the example; opaque-`creator_ref` identity rules repeat across the contract, steps 1 and 8; and 'never enters the SQS' recurs three times. Mostly efficient but could be tightened by stating each invariant once — not a 2 because there is no concept-explanation filler, only skill-specific repetition. | 3 / 5 |
Actionability | Concrete guidance throughout for an instruction-only skill: exact item states 'Pass/Partial/Fail/Unknown/N/A', an executable command ('python3 "${CLAUDE_PLUGIN_ROOT}/scripts/connectors/youtube.py" videos @handle --limit 10'), exact save paths ('memory/influencer/fit-scorer/YYYY-MM-DD-<topic>.md'), NEEDS_INPUT handling, and template tables in references/scoring-templates.md (verified to exist). Not a 5 because key scoring substance (cohort benchmarks, 'declared rule') is deferred to star-benchmark.md and other files that do not resolve within this bundle. | 4 / 5 |
Workflow Clarity | Eight clearly sequenced steps with explicit validation checkpoints and recovery loops: missing context returns NEEDS_INPUT 'name the missing fields, and preserve the supplied identity unchanged for resume'; 'Unknown means applicable evidence is missing and prevents a complete Suitability read'; a Fail on STAR-S2/STAR-S6 operationally holds outreach; saves require exact authorization; and a 'Done when' checklist plus a max-depth-3 termination rule close the loop. Not below 5 because checkpoints, error recovery, and a completion checklist are all explicitly written. | 5 / 5 |
Progressive Disclosure | Good structure: well-labeled sections, and the one in-bundle reference (references/scoring-templates.md) is clearly signaled with a content description and is one level deep. Not a 5 because many referenced paths (star-benchmark.md, skill-contract.md, state-model.md, CONNECTORS.md, sibling SKILL.md files) resolve outside this bundle where they cannot be verified, and sibling links are listed redundantly across 'Reference Materials', 'Next Best Skill', and 'Related Skills'. | 4 / 5 |
Total | 16 / 20 Passed |