<!-- plugin/skills/qa-hardening/SKILL.md -->

---

name: qa-hardening description: This skill should be used when the user asks for "QA review", "production readiness", "hardening pass", "test coverage review", "error handling audit", "pre-release check", or wants to check code quality before release. argument-hint: "[area to review]" disable-model-invocation: true

QA Hardening

Requires: Claude Code Agent Teams feature.

Production readiness review with 4 specialists: test health auditor, error handling auditor, security auditor, and edge case hunter.

Arguments: $ARGUMENTS

Instructions

1. Parse arguments (optional):

   • --members hana,kali -> Only spawn these specific agents (by first name)
   • No arguments -> spawn all 4 agents
   • Any other text -> use as the context/focus for the review

2. Determine context: The area to review, specific concerns, or scope of analysis. If no context is obvious, ask the user what they'd like hardened.

3. Launch the team: Call the Mira launch MCP tool to get agent specs:

   launch(team="qa-hardening-team", scope=user_context, members="hana,kali" or omit for all)
   

   The members parameter is only needed if the user passed --members. The scope parameter should describe the review focus.

4. Create the team:

   TeamCreate(team_name=result.data.suggested_team_id)
   

5. Create and assign tasks: For each agent in result.data.agents:

   TaskCreate(subject=agent.task_subject, description=agent.task_description)
   TaskUpdate(taskId=id, owner=agent.name, status="in_progress")
   

6. Spawn agents: For each agent in result.data.agents, use the Task tool:

   Task(
     subagent_type="general-purpose",
     name=agent.name,
     model=agent.model,
     team_name=result.data.suggested_team_id,
     prompt=agent.prompt + "\n\n## Context\n\n" + user_context,
     run_in_background=true
   )
   

   Spawn all agents in parallel (multiple Task calls in one message).

   IMPORTANT: Do NOT use mode="bypassPermissions" -- these are read-only discovery agents. IMPORTANT: Always pass model="sonnet" to the Task tool. This ensures read-only agents use a cost-efficient model.

7. Wait for findings: All agents will send their findings via SendMessage when complete.

8. Synthesize findings: Combine all findings into a prioritized hardening backlog:

   • Critical -- Must fix before release (panics in production paths, security holes, data loss)
   • High -- Should fix before release (poor error messages, resource leaks, missing validation)
   • Medium -- Fix soon after release (coverage gaps, edge cases, docs drift)
   • Low -- Polish (naming consistency, minor UX, nice-to-have tests)
   • Deferred -- Needs design discussion (architectural changes, large refactors)

   Cross-reference findings -- when multiple agents flag the same area, elevate priority.

9. Cleanup: Send shutdown_request to each teammate, then call TeamDelete.

Want findings implemented?

After presenting the hardening backlog, ask the user if they want fixes implemented. If yes, use launch(team="implement-team", scope=approved_items) to get implementation agent specs, then spawn Kai for planning, parallel fixers for execution, and Rio for verification.

Examples

/mira:qa-hardening
-> Prompts for what to review, then spawns all 4 agents

/mira:qa-hardening Review the recipe system
-> All 4 agents review the recipe code for production readiness

/mira:qa-hardening --members kali,orin
-> Only spawns Kali (security) and Orin (error-handling)

Agent Roles

Name	Role	Focus
Hana	Test Health Auditor	Test suite health, coverage gaps, build quality
Orin	Error Handling Auditor	Panic paths, error messages, recovery
Kali	Security Auditor	Input validation, data exposure, auth bypass
Zara	Edge Case Hunter	Boundary conditions, concurrency, resource exhaustion