Conciseness

The body is dense and operational (tables for steps, when/when-not, and error handling) with no concept explanations Claude already knows, matching 'efficient'. Minor trimmable redundancy: the 'NEVER invoke {"skill": "azure-deploy"}' rule is stated three times (deploy-recovery note, Step 8, Workflow header note), and the Post-scaffold transition paragraph is wordy — this keeps it below the 'every token earns its place' anchor.

Actionability

Fully concrete instruction-level guidance: named session artifacts (prereq-output.json, prepare-plan.json, scaffold-manifest.json, deploy-result.json), exact skill invocations ('{"skill": "azure-app-onboard-prereq"}'), machine-checkable halt conditions ('overallHealth: "blocked" OR routeToSkill set'), the exact approval prompt text, and per-step required file reads — copy-paste-ready operational direction covering the common cases.

Workflow Clarity

A 10-step sequenced pipeline with explicit validation checkpoints (Scaffold Gate before generating files, Deploy Gate before any deployment, self-review in Step 7, health-check in Step 9), explicit halt conditions, a checklist (deploy-checklist.md), and an error-handling table with a retry/skip/abort feedback loop — matching the level-5 anchor for destructive/batch operations.

Progressive Disclosure

Good structure with clearly signaled, one-level-deep references that all exist in references/ (pipeline-rules.md, session-protocol.md, intent-gathering.md, approval-gates.md, handoff-protocol.md, mcp-tool-reference.md, iac-resources.md), matching the 'mostly clear, minor organization gaps' anchor. The gap: 'prepare/SKILL.md', 'scaffold/SKILL.md', and 'deploy/SKILL.md' are marked MUST READ in Steps 5/7/9 but are absent from the bundle, leaving broken navigation for three core steps.

Specificity

Lists multiple concrete actions covering the full pipeline ('Analyzes your app, auto-detects the right Azure services, scaffolds infrastructure code, and deploys', 'cost estimates and pre-deploy approval', 'moving existing apps to Azure without rewriting'), which matches the level-5 anchor. However, the description uses second-person voice ('Analyzes your app', 'tailored to your app'), which per the judging guidelines reduces the specificity score by 1.

Completeness

Explicitly answers both 'what' (orchestrator from idea/existing app to running Azure deployment with analysis, service detection, scaffolding, and deploy) and 'when' (a dedicated 'WHEN:' clause with concrete trigger phrases), plus a 'DO NOT USE FOR' clause with redirects — matching the level-5 anchor exactly.

Trigger Term Quality

The 'WHEN:' clause lists ~16 natural phrases users would actually say, including synonyms and variations ('deploy my app to the cloud', 'deploy my new app to Azure', 'one-click deploy', 'migrate my app to Azure', 'cost to run', 'no code yet'), matching the comprehensive-coverage anchor.

Distinctiveness Conflict Risk

A 'DO NOT USE FOR' clause explicitly disambiguates the closest siblings ('use azd for deployment (use azure-deploy)', 'optimizing existing costs (use cost-optimization)', 'code readiness checks only (use azure-app-onboard-prereq)'), giving a clear niche. Broad catch-all triggers like 'help me get started', 'build an app', and 'starter project' create minor overlap risk with general app-building skills, so it fits the 'mostly distinct; minor overlap risk' anchor rather than level 5.