Conciseness

The prerequisite gate is stated four times — the AUTHORITATIVE GUIDANCE block ("azure-validate skill MUST be invoked and completed with status Validated BEFORE executing"), the STOP block, Rules 1-2 ("must exist with status Validated"), and Step 1 — and failure warnings repeat ("Skipping steps causes deployment failures", "deployments will fail"). Mostly efficient (tight tables, lean 88-line body) but the duplicated gating prose could be consolidated into one statement plus the Step 1 check, which fits the 'could be tightened' anchor rather than the 'several padded sections' of score 2.

Actionability

Steps carry concrete, executable specifics where it matters: "run azd provision --no-prompt, then verify AcrPull role has propagated" (Step 4), "verify status = Validated AND Validation Proof section is populated" (Step 1), and the URL-format rule gives an explicit https:// example; every step also names the file that holds the actual commands. It sits between anchors because core deploy execution is fully delegated to the recipe files rather than shown, leaving minor gaps like Step 5's bare "Follow recipe steps".

Workflow Clarity

The 10-step table gives a clear sequence with explicit validation checkpoints at the front (plan status + Validation Proof), middle (RBAC propagation gate before image deploy), and end (verify success, live role verification), plus a dedicated error-handling step routing to each recipe's errors.md and a mandatory pre-deploy checklist. This matches the top anchor: explicit validation steps, feedback loops for error recovery, and a checklist for a complex, partly destructive process.

Progressive Disclosure

The body is a lean overview with well-signaled references, and every referenced path resolves to a real bundle file (including the deep anchor into pre-deploy-checklist.md's RBAC section). It stops short of the 5 anchor because the recipe content chains two hops from the top (SKILL.md → recipes/README.md → azd/README.md → errors.md) and Step 7's reference column is "—" with no direct link to the recipe's errors.md — minor organization gaps rather than the inlining or buried references of score 3.

Specificity

"This skill runs azd up, azd deploy, terraform apply, and az deployment commands with built-in error recovery" lists several concrete, specific actions with clearly defined inputs ("existing .azure/deployment-plan.md and infrastructure files"). It falls just short of the 5 anchor because post-deployment verification and reporting — actual steps of this skill — are not mentioned, a minor coverage gap.

Completeness

Both questions are explicitly answered: the 'what' ("Execute Azure deployments for ALREADY-PREPARED applications... runs azd up, azd deploy, terraform apply, and az deployment commands") and the 'when' (a concrete WHEN list of trigger phrases), plus an explicit DO NOT USE WHEN list. This squarely fits the anchor for clearly and explicitly answering both what and when with concrete trigger phrases.

Trigger Term Quality

The WHEN clause covers a comprehensive set of natural phrases users would actually say — "push to production", "go live", "ship it", "launch on Azure", "publish to Azure" — alongside precise technical triggers like "run azd up" and "terraform apply". It clearly matches the comprehensive-coverage anchor rather than score 4, since no common synonym or variation is obviously missing.

Distinctiveness Conflict Risk

The description carves out a clear niche (deploying already-prepared apps) and actively resolves the main conflict by stating "DO NOT use this skill when the user asks to CREATE a new application — use azure-prepare instead" with matching anti-triggers ("create and deploy", "build and deploy"). It matches the clear-niche-with-minimal-conflict-risk anchor; the sibling-skill references reduce rather than increase overlap.