Conciseness

The body is dense and assumes competence (no explanation of what diffs or code reviews are), but the --reviewer mandate is repeated in four places (Pipeline comment, 'What gets rendered', Hard rule 1, Forcing question 1) and the WCAG color+icon+aria point appears three times. Matches 'efficient; minor instances... that could be trimmed'; not 5 because of that redundancy, not 3 because there is no real over-explanation of known concepts.

Actionability

The Pipeline section gives fully executable, copy-paste-ready commands with concrete flag values ('--reviewer "Jane Doe" --title "PR #123: Add retry logic"'), full script paths, exit codes for refusals (3, 4), and a worked custom-severity example ('--severity-convention "critical,important,suggestion,nit"'). Matches the anchor for copy-paste ready commands covering common cases.

Workflow Clarity

The three-step pipeline is clearly sequenced with named intermediate artifacts (hunks.json → annotations.json → review.html) plus a diagram, and refusal gates (exit 3/4) act as checkpoints. Not 5 because there is no explicit verification of the rendered output or a validate-and-retry feedback loop (e.g., what to check in review.html before finishing). Not capped at 3 since the operation is non-destructive single-file generation.

Progressive Disclosure

The body is a proper overview (~87 lines) and the three references/ files (diff_rendering_canon, severity_coding, pr_annotation_ux — all verified to exist) are one level deep, but they are only named parenthetically in the final References section with no signal for when to consult each during the workflow. Matches 'references mostly clear; minor organization gaps'; not 5 for that missing per-reference navigation cue.

Specificity

Lists multiple concrete actions with comprehensive coverage: 'Converts a markdown PR writeup or code review... into a single-file 2-column HTML review — unified-diff on the left, severity-tagged annotation cards on the right, top jump-nav listing every finding, mandatory named reviewer footer.' Matches the anchor for multiple specific concrete actions; not 4 because no meaningful capability is left out.

Completeness

Explicitly answers both: 'what' (the full conversion capability described above) and 'when' ('Triggers when the markdown-html-orchestrator classifies an input as REVIEW, or when invoked directly via /cs-md-review. Use after orchestrator routing.'). Concrete trigger phrases are present, matching the anchor 5 example; no cap applies since an explicit trigger clause exists.

Trigger Term Quality

Good natural keyword coverage: 'markdown PR writeup', 'code review', '```diff fenced blocks', '[!BLOCKER]/[!MAJOR]/[!MINOR]/[!NIT] callouts', '/cs-md-review'. Not 5 because trigger phrasing leans on internal routing ('markdown-html-orchestrator classifies an input as REVIEW') rather than natural user variations like 'review comments' or 'PR feedback'; not 3 because the domain terms users would actually say are present and specific.

Distinctiveness Conflict Risk

Clear niche (diff hunks + severity callouts → HTML review) with explicit de-confliction: 'refuses if no diff hunks present (route to md-document instead)'. The input signature (diff blocks + severity-tagged GFM callouts) is highly unlikely to trigger the wrong skill; matches anchor 5.