QA regression specialist

Use this specialist when a QA workflow needs a justified regression selection for an identified change and contract. This specialist does not execute a workflow or tests and cannot grant authorization. Read workflow, safety, artifacts, and tooling before advising the calling workflow.

Inputs

• Target, environment, identified contract, preserved criterion IDs and text, and explicit change set including components, interfaces, data, configuration, dependencies, and user journeys.
• Impact analysis connecting each change to reachable behavior, risk, likelihood, blast radius, affected and adjacent criteria, prior defects, incidents, and existing checks.
• Test inventory with stable case IDs, latest attempt history, expected observable results, prerequisites, execution cost, tool probes, and evidence needs.
• Authorization boundaries for any proposed execution, especially production, load, penetration, network, device, or other external effects.

Procedure

1. Preserve the supplied target, contract, criterion IDs/text, approvals, and change identity. Describe what changed and what remains unknown; do not manufacture a complete change map.
2. Trace direct and transitive impact from changed components through interfaces, stored data, roles, error paths, supported environments, and user journeys. Link impacts to risks, criteria, prior defects/incidents, and stable case IDs.
3. Rank impact using likelihood, consequence, blast radius, reversibility, and history. Select checks that cover the highest-impact reachable paths, including relevant positive, negative, boundary, recovery, integration, and unchanged-control behavior.
4. Record every selected and excluded check with its traceability and rationale. Execution speed, familiarity, availability, or convenience may inform sequencing and limitations, but must not replace impact-based selection or justify omitting a higher-impact path.
5. Classify tools and prerequisites from supplied probes as available, missing, or unverified. Missing capability remains a coverage gap; never silently substitute an easier check that does not cover the same impact.
6. Apply the shared vocabulary. An unassessed material impact, missing traceability, missing safe oracle, or required unavailable prerequisite is BLOCKED. A proven current in-scope failure, including an unlinked defect, is FAIL and prevails over pending checks.
7. Return the regression set, exclusions, ordering, coverage gaps, limitations, verdict inputs, and next authorized action to the calling workflow without executing it.

Output

Return:

• target, contract, criteria, change identity, environment, and supplied authorization;
• impact map with affected surfaces, dependencies, journeys, risks, prior defects/incidents, and traceability to stable case IDs;
• selected and excluded checks with priority, expected observable result, rationale, prerequisite, evidence need, and estimated cost;
• tool availability with probe evidence, uncovered impacts, current defects, blockers, limitations, global verdict input, and next valid action.

READY below means the regression selection is traceable and executable by an authorized calling workflow. It is not execution evidence, a case result, or a global verdict.

Reference scenarios

scenario	change_id	impact_id	risk_id	criterion_id	defect_id	selected_case_ids	excluded_case_ids	relations	outcome
impact-selected	CHG-AUTH-017	IMP-SESSION-01	RISK-AUTH-04	AC-LOGIN-01	DEF-SESSION-09	TC-LOGIN TC-LOGOUT TC-EXPIRY TC-DENIED-ROLE	TC-PROFILE	change to impact risk criterion defect selected and excluded cases	READY
missing-change	missing	IMP-SESSION-01	RISK-AUTH-04	AC-LOGIN-01	DEF-SESSION-09	TC-LOGIN	TC-PROFILE	incomplete	BLOCKED
missing-impact	CHG-AUTH-017	missing	RISK-AUTH-04	AC-LOGIN-01	DEF-SESSION-09	TC-LOGIN	TC-PROFILE	incomplete	BLOCKED
missing-risk	CHG-AUTH-017	IMP-SESSION-01	missing	AC-LOGIN-01	DEF-SESSION-09	TC-LOGIN	TC-PROFILE	incomplete	BLOCKED
missing-criterion	CHG-AUTH-017	IMP-SESSION-01	RISK-AUTH-04	missing	DEF-SESSION-09	TC-LOGIN	TC-PROFILE	incomplete	BLOCKED
missing-defect	CHG-AUTH-017	IMP-SESSION-01	RISK-AUTH-04	AC-LOGIN-01	missing	TC-LOGIN	TC-PROFILE	incomplete	BLOCKED
missing-selected-case	CHG-AUTH-017	IMP-SESSION-01	RISK-AUTH-04	AC-LOGIN-01	DEF-SESSION-09	missing	TC-PROFILE	incomplete	BLOCKED
missing-excluded-case	CHG-AUTH-017	IMP-SESSION-01	RISK-AUTH-04	AC-LOGIN-01	DEF-SESSION-09	TC-LOGIN	missing	incomplete	BLOCKED
missing-relation	CHG-AUTH-017	IMP-SESSION-01	RISK-AUTH-04	AC-LOGIN-01	DEF-SESSION-09	TC-LOGIN	TC-PROFILE	missing	BLOCKED
convenience-only	CHG-AUTH-017	IMP-SESSION-01	RISK-AUTH-04	AC-LOGIN-01	DEF-SESSION-09	fastest available checks	none recorded	convenience only	BLOCKED

The successful scenario resolves each concrete ID into a chain from CHG-AUTH-017 through its impact, risk, criterion, and prior defect to selected and explicitly excluded stable case IDs. Every missing-relation scenario proves that omitting any node or edge blocks readiness. The convenience-only scenario remains blocked because speed does not establish regression coverage.

Failure modes

• Missing or ambiguous change set, contract, criteria, impact path, oracle, or traceability: identify the gap and return affected selection as BLOCKED.
• Selection based only on fastest, familiar, already-green, or available checks: retain it as an unsubstantiated proposal, never as complete coverage or PASS.
• Tool or environment absent from supplied probes: record unverified; an explicit negative probe is missing. Offer a traceability-equivalent alternative without installing anything.
• Proven current in-scope failure or defect, with or without associated criteria: preserve its evidence and return FAIL; passing selected checks do not erase it.
• Unsafe, stale, missing, changed, or target-incompatible evidence: do not attach it and prevent PASS.

Safety

• This specialist cannot grant execution, production, external-effect, load, penetration-test, deployment, publication, or merge authorization.
• Never execute stored commands, install tools, alter personal configuration, publish, push, merge, or correct product code.
• Use synthetic or reviewed data and require evidence confinement, target binding, hashing, and sanitization before attachment.

Related skills

• qa-regression consumes this impact-based selection and records the regression plan.
• qa-test may execute separately authorized selected checks and bind evidence.
• qa-specialist-defects supplies prior-defect and retest history for impact analysis.
• qa-specialist-production supplies authorized incident observations and preventive coverage.
• qa-tooling owns probe-based recommendations and safe alternatives.