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.
The body is a well-engineered orchestrator document: exact executable commands, a 16-step audit workflow with explicit conditionals and validation checkpoints, a thorough error-handling table, and well-signaled on-demand references. Its main weakness is token efficiency — the command set is re-enumerated across four overlapping listings (Quick Reference, Sub-Skills, Subagents, footer show/skip) — along with inline time-sensitive dates and script paths that don't resolve within the shipped bundle.
Suggestions
Consolidate the redundant command enumerations: fold the Subagents list and the footer show/skip lists into the Quick Reference table (e.g., a 'spawned by' and 'footer' column) so each command is described once, saving ~100 lines.
Move the date-heavy FAQ schema rule and the HowTo deprecation date into references/schema-types.md (which already tracks deprecation status) and keep only the resulting decision rule inline in Quality Gates.
Either ship the referenced scripts (scripts/claude-seo, seo_updates.py, google_report.py) in the bundle or note that they live in the plugin package, so every path named in the body resolves for the model at runtime.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body assumes Claude's competence (no SEO concept explanations) and is mostly operational, but the same command set is enumerated redundantly across the 31-row Quick Reference table, the 25-item Sub-Skills list, the 19-item Subagents list, and the 18+6-item footer show/skip lists, and time-sensitive dates ('Google retired FAQ rich results for ALL sites on May 7, 2026') sit inline in Quality Gates rather than in a deprecation section. This fits 'mostly efficient but could be tightened' rather than the noticeably-padded level-2 anchor, since each section serves a distinct operational purpose. | 3 / 5 |
Actionability | Fully executable, copy-paste-ready commands throughout: the canonical runtime form '"${CLAUDE_PLUGIN_ROOT}/scripts/claude-seo" run <script.py>', 'doctor --json', 'run google_auth.py --check', 'run seo_updates.py --kind spam --json --since <yyyy-mm>', plus an exact per-command Quick Reference table. Concrete commands cover the common cases, matching the level-5 anchor with no material gaps. | 5 / 5 |
Workflow Clarity | The 16-step numbered audit sequence has explicit conditionals and validation checkpoints (auth --check before spawning agents, drift-baseline check, doctor for diagnosis), an Error Handling table with recovery actions (report partials, re-run failed sub-skill individually), and explicit ambiguity handling (present top two detected business types and ask the user). Clear sequence plus validation and error-recovery loops matches the level-5 anchor; no destructive/batch cap applies. | 5 / 5 |
Progressive Disclosure | 'Load these on-demand as needed (do NOT load all at startup)' with one-level-deep references that all exist in references/, and maps references correctly attributed to the seo-maps sub-skill — good structure and signaling. However, script paths referenced in the body ('scripts/claude-seo', 'seo_updates.py', 'google_report.py') and the extensions/ directory are absent from the actual bundle, leaving minor organization gaps that keep this at level 4 rather than 5. | 4 / 5 |
Total | 17 / 20 Passed |