Conciseness

The steps table, rules, and reference listings are lean, but the opening prerequisite block pads heavily with repetition — the same requirement is restated in the 'AUTHORITATIVE GUIDANCE' blockquote, the '⛔ STOP' check, Rules 1-2, Step 1, and the 'VALIDATION PROOF CHECK' note ('DO NOT ASSUME', 'DO NOT SKIP', 'Skipping steps causes deployment failures'). The warning content could be stated once, so this sits at 'mostly efficient with some unnecessary explanation' rather than a lower score since the operational content itself is tight.

Actionability

Gives one concrete executable command ('run `azd provision --no-prompt`, then verify `AcrPull` role has propagated'), a precise steps table with file paths, a named MCP tool table with specific tool identifiers like `mcp_azure_mcp_azd` and `azure__role`, and an explicit URL-formatting rule. It falls short of fully executable because the core deploy execution is delegated to recipe READMEs rather than including even one representative deploy command inline.

Workflow Clarity

A 10-step table clearly sequences the workflow with explicit validation checkpoints (plan status = `Validated` with populated Validation Proof, mandatory pre-deploy checklist, RBAC health check, verify success with endpoint accessibility, live role verification), plus a dedicated error-recovery step routing to each recipe's `errors.md`. This matches the anchor for clear sequence with explicit validation steps and feedback loops for error recovery.

Progressive Disclosure

The body is a clean overview: every referenced path (pre-deploy-checklist.md, global-rules.md, recipes/README.md, recipes/azd/post-deployment.md, recipes/azd/verify.md, live-role-verification.md, all five sdk/ files, troubleshooting.md) exists on disk, links are one level deep and clearly labeled with purpose, and detailed content is properly split out. The deep-link anchor `#container-apps--acr--pre-deploy-rbac-health-check` resolves to a real heading in the checklist. Navigation is easy and well-signaled.

Specificity

Multiple specific concrete actions in third person ('runs azd up, azd deploy, terraform apply, and az deployment commands with built-in error recovery') with comprehensive coverage of the deployment toolchain. It stays at 5 rather than 4 because the actions span every infrastructure approach rather than having minor coverage gaps.

Completeness

Explicitly answers what ('Execute Azure deployments for ALREADY-PREPARED applications... runs azd up, azd deploy, terraform apply, and az deployment commands'), when (a WHEN list of nine concrete trigger phrases), and when-not (DO NOT USE WHEN list with redirects to azure-prepare), plus prerequisites. This exceeds the 5 anchor, which requires concrete trigger phrases for both what and when.

Trigger Term Quality

Comprehensive natural-phrase coverage with synonyms across registers: technical ('run azd up', 'bicep deploy', 'terraform apply'), colloquial ('push to production', 'go live', 'ship it', 'push to cloud'), and destination-oriented ('publish to Azure', 'launch on Azure'). All are phrases a user would naturally say, matching the 5 anchor's standard of comprehensive synonym coverage.

Distinctiveness Conflict Risk

A clear niche (deployment execution only, for already-prepared apps) with explicit disambiguation: 'DO NOT use this skill when the user asks to CREATE a new application — use azure-prepare instead' and negative triggers like 'create and deploy', 'build and deploy', 'set up infrastructure' routed to azure-prepare. The create-vs-deploy boundary is the exact overlap risk for this skill family and it is handled explicitly, so conflict risk is minimal.