Content
88%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.
An exemplary orchestration skill body: per-command numbered workflows, exact file paths with accurate on-disk counts, a smart selective-loading policy for the 21-prompt optimize stage, and a concrete error-recovery table. The only deductions are minor — a duplicated load instruction/attribution text and a sync-script reference that does not resolve within the provided bundle.
Suggestions
State the flow-framework.md load-on-activation instruction once — it currently appears in both the 'Runtime context' note and the 'Reference Files' section.
Resolve or qualify the sync reference: the body points to ${CLAUDE_PLUGIN_ROOT}/scripts/claude-seo (sync_flow.py) but no scripts/ directory ships in this skill bundle; either bundle the script or note that it lives at the plugin root so the reference is verifiable.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is a tight, table-driven orchestration guide with no explanations of concepts Claude already knows and an explicit anti-dump policy ("The optimize stage has 21 prompts. Dumping all 21 is noise"). Minor trimming is possible: the instruction to load references/flow-framework.md on every activation is stated twice ("Runtime context" and "Reference Files"), and the attribution text appears both inline and again verbatim in its own section — fitting the 'efficient; minor instances that could be trimmed' anchor rather than the every-token-earns-its-place anchor at 5. | 4 / 5 |
Actionability | Guidance is fully executable: exact per-subcommand numbered sequences, exact file paths per stage, a copy-paste sync invocation ("${CLAUDE_PLUGIN_ROOT}/scripts/claude-seo" run sync_flow.py), concrete cross-reference syntax (e.g., "see /seo cluster <seed-keyword>"), and a priority-ordered prompt-selection rule (industry vertical → prior skill output → URL signals). For an orchestration skill this is copy-paste-ready command guidance covering all cases, matching the top anchor. | 5 / 5 |
Workflow Clarity | Every sub-command has a clearly sequenced numbered workflow, and the Error Handling table provides explicit recovery loops (missing reference → "Run /seo flow sync"; network error → check "gh api rate_limit"; auth error → "Run gh auth login then retry"). The optimize flow adds a checkpoint constraint ("Always surface exactly 2-3 prompts. State which prompts you chose and why"), matching the anchor for clear sequences with explicit validation steps and feedback loops for error recovery. | 5 / 5 |
Progressive Disclosure | The body is overview-only with a dedicated Reference Files section giving per-reference load conditions ("load on every /seo flow activation", "load when citing studies", "load selectively") and an explicit on-demand policy; all referenced paths and prompt counts verified against the actual bundle (find=5, leverage=1, optimize=21, win=3, local=11, total 41 — all exact). The one gap: the sync workflow references ${CLAUDE_PLUGIN_ROOT}/scripts/claude-seo and sync_flow.py, but no scripts/ directory exists in the skill bundle, leaving that referenced path unverifiable — a minor organization gap that keeps it below the top anchor. | 4 / 5 |
Total | 18 / 20 Passed |