Release-Coverage Advisor

Before you can ask "is this release ready?" you have to know what a release should even test. The Release-Coverage Advisor answers that: point it at a project (a release) and it tells you the test coverage each of its surfaces should have — and which of those checks are actually buildable right now.

It is read-only and deterministic. It creates nothing, runs nothing, and produces no verdict — it's purely the planning step, upstream of building tests and of readiness scoring.

What it does

For every surface under the project — an (app × environment × target) — it recommends coverage keyed by the target's type:

Surface	Recommended coverage
REST service with an OpenAPI spec	API contract · performance smoke · security sweep
REST service without a spec	(blocked — add a swagger path)
Web app	UI flow · crawl
Network / IP / hostname	security sweep

Each recommended item comes back marked ready (buildable now) or with human-readable blockers — e.g. "no OpenAPI spec", "app requires auth but no credential resolves in this environment". Apps in the project that have no target to test show up under gaps.

What makes a REST surface "ready" for API contract testing: the application's swaggerPath resolves to a reachable OpenAPI spec (the advisor uses the same resolver the generator does). Set the swagger path on the application — not just the target — so API_CONTRACT reports ready.

Using it

MCP:

recommend_release_coverage(project="User Management Platform")
recommend_release_coverage(project="User Management Platform", environment="staging")

REST:

GET /api/projects/{projectId}/release-coverage
GET /api/projects/{projectId}/release-coverage?environmentId={envId}

Example response (trimmed):

{
  "projectName": "User Management Platform",
  "surfaces": [
    {
      "app": "user-service",
      "environmentName": "development",
      "targetType": "REST_SERVICE",
      "recommended": [
        { "kind": "API_CONTRACT",  "ready": true,  "blockers": [] },
        { "kind": "PERF_SMOKE",    "ready": true,  "blockers": [] },
        { "kind": "SECURITY_SWEEP","ready": true,  "blockers": [] }
      ]
    }
  ],
  "gaps": [],
  "summary": "1 surface(s) · 3 recommended check(s) · 0 blocked · 0 gap(s)"
}

From plan to tests

A ready item maps directly to the generator that builds it:

Recommended kind	Build it with
API_CONTRACT	create_scenario_from_swagger
PERF_SMOKE	convert_api_to_performance_scenario
UI_FLOW / CRAWL	crawl_by_target → then a UI test
SECURITY_SWEEP	a security scan

Once the recommended coverage is built and run, use readiness scoring (see How Scoring Works) to judge whether the release is ready to ship.

note

The advisor is a snapshot of what the release should test, taken when you ask. If the app's surfaces change later, re-run it to reconcile — it won't silently mutate a plan underneath you.