AgentSkillsCN

Automated Testing

自动化测试

SKILL.md
--- frontmatter
skill_id: CICD-AUTO-TEST
version: 1.0.0
last_updated: 2026-01-04
applies_to: [Class A, Class B, Class C]
jurisdiction: [Global]
prerequisites: [TEST-UNIT, TEST-INTEGRATION]

Automated Testing in CI/CD

Purpose

Define how to automate tests (unit, integration, system/HIL where feasible) in CI/CD pipelines for medical devices with gating and evidence capture.

When to Apply

  • •Setting up or updating test stages in CI/CD.

Requirements (testable)

  1. •Scope: Include unit, integration, and where possible system/HIL or simulators; prioritize safety-related modules. Rationale: coverage of risk areas.
  2. •Gating: Fail pipeline on test failures; no ignored failures for safety code. Rationale: prevent regressions.
  3. •Artifacts: Store test results (JUnit/XML), logs, coverage, and evidence; retain per release. Rationale: auditability.
  4. •Flake Management: Detect and address flaky tests; quarantine with issue filed and timeline; do not ignore. Rationale: trustworthiness.
  5. •Hardware Tests: If hardware required, provide conditional stages and fallbacks (sim); schedule regular hardware runs (nightly). Rationale: practicality with coverage.

Recommended Practices

  • •Tag tests by type/priority; run critical ones on every commit.
  • •Use retries sparingly and only for known flaky, with tracking.
  • •Parallelize to keep feedback fast.
  • •Use containerized or pinned environments.

Patterns

CI job snippet:

yaml
test:
  script: [ "ctest --output-on-failure" ]
  artifacts:
    when: always
    paths: [ "build/test-results.xml", "coverage/" ]

Flake tracking entry:

yaml
test_id: TEST-INT-210
status: flaky
issue: QA-1234
expires: 2026-02-01

Anti-Patterns (risks)

  • •Allowing tests to fail without blocking or tracking -> risk: regressions.
  • •No artifacts/logs retained -> risk: unverifiable results.
  • •Silent retries masking real failures -> risk: shipping defects.

Verification Checklist

  • • Unit/integration (and sim/HIL if available) tests run in CI.
  • • Failures gate merges; no ignored failures for safety code.
  • • Results/logs/coverage stored as artifacts per run/release.
  • • Flaky tests tracked with issues and timelines.
  • • Hardware-dependent tests scheduled; sim fallback provided.

Traceability

  • •Link CI test runs to test IDs (TEST-###) and requirements; store artifacts with release metadata.

References

  • •IEC 62304 verification expectations.
  • •FDA/MDR submission evidence needs.

Changelog

  • •1.0.0 (2026-01-04): Initial automated testing skill with gating and artifact capture.

Audit History

  • •2026-01-04: Audit performed. Verified:
    • •IEC 62304 verification expectations correctly referenced
    • •CI/CD patterns follow industry best practices for regulated environments