feat(sdlc): add test-authoring skill and agent with supporting sub-sk…

[AnastasiiaHarkot]

⚠️ REQUIRES HUMAN REVIEW - Trust Files Changed

This PR modifies agent trust-path files and requires human review before AI agents can run on it.

Modified trust files:

• .claude/skills/test-authoring/SKILL.md (8 edits in trust-path)

---

Introduces the test-authoring Claude skill (SKILL.md), a companion agent definition (agents/test-authoring/agent.yml), and a set of reusable sub-skills (notify-slack, qa-soul, report-status, tool-conventions, traceability-lookup) to support structured manual test-case authoring and traceability-matrix updates.

📋 Latest Changes (NEW)

Commit 1: Add triggers and dependencies

• ✅ Add frontmatter with triggers and dependencies fields
• ✅ Add "Overview" section describing skill purpose
• ✅ Add "When to use" section (automatic/manual/as action scenarios)
• ✅ Add "Triggers and dependencies" section documenting 5-step pipeline
• ✅ Add "Required tools" section

Commit 2: Replace requirements-check with GitHub Actions

• ✅ Replace requirements-check agent with github-actions in triggers
• ✅ Update trigger gate to read GitHub Actions environment variables
• ✅ Add GitHub Actions inputs specification:
  • CHANGED_SPECS: JSON array of changed spec file paths
  • CHANGE_TYPE: Type of change (new/modified/removed)
  • REQUIREMENT_IDS: Extracted requirement IDs from specs
• ✅ Update procedure to parse data from GitHub Actions environment
• ✅ Update pipeline: spec changes → spec-validation → test-authoring → code-review → test-automation

🔒 Security Review Needed

Why human review is required:

• Modified .claude/skills/test-authoring/SKILL.md (agent configuration)
• Protection against config-poisoning / prompt-injection attacks
• AI agents cannot auto-run on trust-file changes per M2 security policy

Changes are safe because:

• ✅ Only documentation and trigger configuration updates
• ✅ No executable code changes
• ✅ No changes to permissions or allowed-tools
• ✅ No prompt injection attempts
• ✅ Legitimate refactoring to support GitHub Actions workflow

👥 Reviewers Please Verify

• No malicious prompt injections in SKILL.md
• Configuration changes are intentional and documented
• No security bypasses or permission escalations
• Changes align with stated purpose
• GitHub Actions integration is properly specified

📝 Recent Commits

• b34064c67 feat(skills): add triggers and dependencies to test-authoring skill
• 78c6653e8 refactor(skills): replace requirements-check with GitHub Actions triggers

---

Note: Once human review is complete and approved, this PR can be merged. AI agents will remain disabled until trust-file changes are verified.

[AnastasiiaHarkot]

🔔 Update: Trust Files Modified - Human Review Required

Hi @Alexander-Kezik @PolinaGurinovich97 @valerydluski

I've updated this PR with 2 new commits that modify the test-authoring skill configuration.

⚠️ Security Notice

This PR now modifies agent trust-path files (.claude/skills/test-authoring/SKILL.md), which triggers the M2 security policy:

• AI agents are blocked from running on this PR automatically
• Human review is required before merge

📋 What Changed

Commit 1: Added triggers, dependencies, and pipeline documentation to SKILL.md
Commit 2: Replaced requirements-check agent with GitHub Actions as the trigger mechanism

✅ Safety Verification Needed

Please review that:

• No prompt injection attempts in SKILL.md changes
• Configuration changes are legitimate and documented
• No security bypasses or permission escalations
• GitHub Actions integration is properly specified

📝 Changes Summary

• Replaced requirements-check agent dependency → github-actions trigger
• Added GitHub Actions inputs: CHANGED_SPECS, CHANGE_TYPE, REQUIREMENT_IDS
• Updated trigger gate to read environment variables instead of .state/status/requirements-check.json
• Documented 5-step pipeline: spec changes → spec-validation → test-authoring → code-review → test-automation

All changes are documentation/configuration only - no executable code modified.

Thank you! 🙏