Regression selection agent

Instructions

Purpose

Select and prioritize regression tests based on the impact map and graph dependency relationships. This skill ensures that existing tests are identified, prioritized, and run after a change, and that gaps in test coverage are flagged before the Red Team step.

---

Trigger conditions

Use this skill when:

• Existing behavior must not break (regression risk).
• An impact map is available.
• The change affects shared modules, services, or APIs.
• The system has existing tests that need to be triaged.
• The Test Designer needs to know which existing tests to include.
• A release gate requires documented regression-test coverage.

---

Inputs

- .agile-v/phases/01-impact/impact_map.md       required
- .agile-v/phases/01-impact/affected_components.json required
- test inventory (discovered from test/ directory or graph test nodes)   required
- .agile-v/phases/00-understanding/normalized_graph.json optional but preferred
- previous test results (if available)          optional

---

Outputs

All outputs are written to canonical .agile-v/phases/03-regression/.

required_regression_tests.md
selected_tests.json
regression_coverage_rationale.md
missing_regression_tests.md

---

Required behavior

Step 1: Discover existing tests

1. Search the repository for test files (.spec.ts, .test.ts, _test.py, test_*.py, etc.).
2. If a normalized graph is available, use test nodes from the graph.
3. Build a test inventory: test_path → linked_components (from imports and graph edges).

Step 2: Select tests linked to impacted components

For each test:

1. Check whether it imports, references, or tests any affected component.
2. Prioritize tests that directly test affected entry points.
3. Assign priority: High (direct), Medium (indirect), Low (general).

Step 3: Identify gaps

For each affected component in affected_components.json:

1. Check whether any selected test covers it.
2. If no test is found, record a gap in missing_regression_tests.md.
3. Suggest a test name and structure for the missing coverage.

Step 4: Build rationale

For each selected test, record:

• Why it was selected.
• Which requirement or risk it covers.
• Which component it links to.

---

Output template: required_regression_tests.md

# Required Regression Tests

| Test | Linked Component | Linked Requirement | Risk Covered | Priority |
|---|---|---|---|---|
| tests/auth/session.test.ts | src/auth/session.ts | REQ-0003 | Session regression | High |

## Missing Regression Tests

| Required behavior | Reason | Suggested test |
|---|---|---|
| Rate limit boundary (5th pass, 6th fail) | No existing test | auth.e2e-spec.ts::boundary_test |

---

Output schema: selected_tests.json

[
  {
    "test_path": "test/auth/auth.e2e-spec.ts",
    "test_name": "should login with valid credentials",
    "linked_component_ids": ["node-042"],
    "linked_requirement_ids": ["REQ-0001"],
    "risk_covered": "Core login path not broken",
    "priority": "high",
    "selection_reason": "Directly tests the affected login endpoint."
  }
]

---

Anti-patterns to avoid

• Do not select all tests in the repository. Select only tests linked to affected components.
• Do not skip gap identification even if the test suite seems comprehensive.
• Do not assign High priority to tests with no clear link to affected components.
• Do not generate new test implementations in this skill (that is test-designer).

Every requirement link must identify a baselined REQ-XXXX revision and baseline; selected tests use test_case -> verifies -> requirement lineage. Apply inherited agile-v-core halt/decision rules and agile-v-aibom manifest capture for materially AI-influenced output at any risk level.

---

Compatible tools

Claude Code, Cursor, Copilot, Codex, Gemini CLI, OpenCode, Cline.