Content
81%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 tightly engineered, contract-style body: lean prose, executable ledger commands, threshold tables, and a well-gated seven-step workflow with genuine validation checkpoints. The gaps are operational rather than structural — several referenced files (scripts/connectors, cross-skill contracts) are missing from the bundle as shipped, and a redundant reference section costs a few tokens.
Suggestions
Ship or stub the referenced scripts/connectors/ledger.py and scripts/connectors/README.md inside the bundle (or remove the zero-dependency ledger loop from the body) so the copy-paste commands actually resolve.
Drop the standalone 'Reference Materials' section and consolidate the three inline citations of tracking-setup-guide / the evidence profile into one place to remove redundancy.
Verify the out-of-bundle cross-references (../../../references/skill-contract.md, ../../../CONNECTORS.md, ../performance-monitor/SKILL.md) resolve from the skill's installed location, since three of them are load-bearing for the handoff and save-results steps.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is dense and lean — contract-style tables, labeled lists, and exact commands with zero explanation of concepts Claude already knows (no SEO primer, no library overviews). Not 5 because of minor redundancy: the 'Reference Materials' section repeats the tracking-setup-guide and evidence-profile links already cited inline, and the performance-monitor link appears in three places. | 4 / 5 |
Actionability | Gives copy-paste-ready commands (`python3 "${CLAUDE_PLUGIN_ROOT}/scripts/connectors/ledger.py" record <domain> --source rank --data '{...}'` plus diff/trend invocations), concrete Quick Start invocations, and a threshold decision table (drop 1-3 → wait 1-2 weeks, etc.). Not 5 because several executable targets are absent from the actual bundle — the referenced scripts/connectors/ledger.py, scripts/connectors/README.md, and the cross-skill evidence-and-cycle-control profile do not resolve — leaving the ledger loop and step 1 partially unexecutable as shipped here. | 4 / 5 |
Workflow Clarity | Seven clearly sequenced steps (setup → record → analyze → features → GEO → competitors → report) wrapped in explicit validation scaffolding: Decision Gates with stop-and-ask options and silent-continue rules, a 'Done when' checklist, evidence labels (Measured/User-provided/Calculated/Estimated/Proxy/Unknown), and NEEDS_REFRESH staleness handling — feedback loops for every ambiguity the workflow can hit. | 5 / 5 |
Progressive Disclosure | The two in-bundle references (references/ranking-analysis-templates.md, references/tracking-setup-guide.md) exist, are one level deep, and are clearly signaled with purpose statements ('complete output templates for all seven steps'; 'tracking setup, root-cause taxonomy, CTR benchmarks'), keeping the body an overview. Not 5 because numerous referenced paths do not resolve within the bundle (../../../references/skill-contract.md cited three times, ../../../CONNECTORS.md, ../performance-monitor/..., scripts/connectors/*), and the duplicate Reference Materials section adds mild navigation noise. | 4 / 5 |
Total | 17 / 20 Passed |