Skip to main content

Published Updated

Create Release

In brief: Create Release is the composite for cutting a release—lock policy, create the catalog entry and test run, run nested QA actions per type, then summarise GATE posture.

When to use​

Use Create Release when it is time to ship: you want one approved plan that creates the release in TestChimp, ensures suite coverage for that cut, optionally runs ExploreChimp / perf / security per policy, and leaves you with a release QA posture summary.

Prefer Run QA after each PR. Prefer Upkeep for portfolio suite maintenance (gaps, cleanup)—not as a substitute for creating a release.

Prompt​

/testchimp create release
/testchimp create release 1.4.0 MINOR
/testchimp create release 2.0.0 MAJOR
/testchimp create release 1.4.1 HOTFIX

Synonyms: /testchimp create release, “create a major|minor|hotfix release…” → catalog id create-release.

Canonical form: /testchimp create release <label> <MAJOR|MINOR|HOTFIX>.

Scope​

Release label and type are locked during the run (from the prompt or interactively). Scope for nested actions follows the locked create-release policy:

  • MAJOR — typically full scenario set
  • MINOR / HOTFIX — typically changed scope (prior release SHA → cut SHA), best-effort

See Workflows intro — Scope for how other workflows resolve omitted scope; this composite is release-identity driven.

Policy​

Default file: plans/knowledge/policies/create-release.policy.md.

Init does not auto-seed a locked create-release policy. The skill ships a seed skeleton with a lock-in banner. On first run the agent proposes recommended defaults (tag convention, per-type matrix), you accept or customize, then the project policy is written without the seed banner and upserted before nested QA starts.

Recommended matrix (playbook defaults—confirm on first lock-in):

TypeScopeTypical actions
MAJORfullupkeep; full suite; ExploreChimp; upkeep-perf + run-perf-tests; security static + DAST
MINORchangedsmart-smoke; ExploreChimp; run-perf-tests; security off
HOTFIXchangedsmart-smoke only

DAST always requires an explicit confirm before scan, even when policy enables it.

Requires a usable connect-to-test-env policy when local suite, ExploreChimp, or DAST needs an environment.

What it does​

Plan → approve → Execute:

  1. Lock label, type, tag convention, scope, and create-release policy (first run: accept/customize defaults).
  2. Create the catalog release (upsert-release with gitCommitSha and optional focus areas) and a named test run titled default on that release.
  3. Ensure suite coverage for the release: prefer CI find/link of existing SmartTests batches when CI is wired; otherwise run full suite or smart-smoke locally with TESTCHIMP_RELEASE / TESTCHIMP_TEST_RUN_ID.
  4. Nest ExploreChimp, perf, and security actions only when the locked policy enables them for that type.
  5. Summarise GATE (get-release detail=GATE) and overall QA posture.

Nested subflows reuse the parent plan. Mutative steps report with policy version and one workflow_execution_id.

See also​

FAQ

When should I run create release vs run QA?

Run /testchimp run QA after each PR. Run /testchimp create release when it is time to ship—create the catalog release, ensure suite coverage for the cut, and summarise GATE posture.

Do I need a create-release policy before the first run?

Not locked in advance. The skill ships a seed skeleton; on first run the agent proposes recommended defaults (tag convention and per-type matrix), you accept or customize, then the project policy is written without the seed banner.

Does create release always run security and perf?

Only when the locked policy enables them for that release type. Recommended defaults: MAJOR includes upkeep, full suite, ExploreChimp, perf, and security; MINOR is lighter; HOTFIX is smart-smoke only. DAST always requires an explicit confirm.

How is suite coverage ensured for the release?

If SmartTests CI is wired, prefer find/link of existing batches for the cut. Otherwise run full suite or smart-smoke locally with TESTCHIMP_RELEASE and TESTCHIMP_TEST_RUN_ID.