User creates issue:
Title: "Login not working"

AI enhances:
- Labels: ["bug", "backend", "authentication", "critical"]
- Priority: "critical" (blocks user access)
- Acceptance Criteria:
  - [ ] Users can log in with valid credentials
  - [ ] Error messages are clear for invalid credentials
  - [ ] Session persists after login

Workflow

Analyze Issue Context
- Parse title for keywords (fix, implement, bug, feature)
- Analyze description for technical details
- Check for urgency markers (critical, urgent, blocking)
- Identify domain from keywords (auth, api, ui, db)
Suggest Labels
- Type: bug, feature, enhancement, documentation, chore
- Domain: backend, frontend, infrastructure, security
- Technology: python, typescript, react, postgresql
- Status: needs-investigation, blocked, ready
- Use RECOMMENDED for clear matches, DEFAULT for inferred
Infer Priority
- Critical: Security vulnerabilities, data loss, complete feature breakage
- High: Major functionality broken, significant user impact
- Medium: Partial functionality affected, workarounds exist
- Low: Minor issues, cosmetic improvements, nice-to-have features
Generate Acceptance Criteria
- Extract implied requirements from description
- Add standard criteria based on issue type:
  - Bugs: Reproduction steps, expected behavior, error handling
  - Features: Happy path, edge cases, error states
- Format as checklist for task tracking

Output Format

{
  "suggested_labels": [
    {"label": "bug", "confidence": "RECOMMENDED", "rationale": "Title contains 'not working'"},
    {"label": "backend", "confidence": "RECOMMENDED", "rationale": "Login is server-side functionality"},
    {"label": "authentication", "confidence": "RECOMMENDED", "rationale": "Login relates to auth"},
    {"label": "critical", "confidence": "RECOMMENDED", "rationale": "Blocks user access"}
  ],
  "suggested_priority": {
    "priority": "critical",
    "confidence": "RECOMMENDED",
    "rationale": "Complete feature breakage preventing user access"
  },
  "acceptance_criteria": [
    {"text": "Users can log in with valid email and password", "confidence": "RECOMMENDED"},
    {"text": "Error messages are displayed for invalid credentials", "confidence": "RECOMMENDED"},
    {"text": "Session persists after successful login", "confidence": "DEFAULT"},
    {"text": "Login form validates input client-side", "confidence": "DEFAULT"}
  ],
  "estimated_complexity": {
    "level": "medium",
    "confidence": "DEFAULT",
    "rationale": "Likely requires debugging auth flow, not full reimplementation"
  }
}

Examples

Example 1: Bug Issue

Input:

{
  "title": "API returns 500 error on user creation",
  "description": "When creating a user with duplicate email, API crashes instead of returning validation error"
}

Output:

{
  "suggested_labels": [
    {"label": "bug", "confidence": "RECOMMENDED"},
    {"label": "backend", "confidence": "RECOMMENDED"},
    {"label": "api", "confidence": "RECOMMENDED"},
    {"label": "validation", "confidence": "RECOMMENDED"}
  ],
  "suggested_priority": {
    "priority": "high",
    "confidence": "RECOMMENDED",
    "rationale": "Server error but has workaround (client-side validation)"
  },
  "acceptance_criteria": [
    {"text": "API returns 422 error with validation message for duplicate email"},
    {"text": "Server does not crash when duplicate email submitted"},
    {"text": "Error response follows RFC 7807 format"}
  ]
}

Example 2: Feature Request

Input:

{
  "title": "Add dark mode support",
  "description": "Users want dark mode for better viewing at night"
}

Output:

{
  "suggested_labels": [
    {"label": "feature", "confidence": "RECOMMENDED"},
    {"label": "frontend", "confidence": "RECOMMENDED"},
    {"label": "ui", "confidence": "RECOMMENDED"}
  ],
  "suggested_priority": {
    "priority": "medium",
    "confidence": "DEFAULT",
    "rationale": "User request but not critical functionality"
  },
  "acceptance_criteria": [
    {"text": "User can toggle dark mode in settings"},
    {"text": "Dark mode persists across sessions"},
    {"text": "All pages support dark mode theme"},
    {"text": "Color contrast meets WCAG AA standards"}
  ]
}

Example 3: Security Issue

Input:

{
  "title": "SQL injection vulnerability in search",
  "description": "Search input is not sanitized, allows arbitrary SQL execution"
}

Output:

{
  "suggested_labels": [
    {"label": "security", "confidence": "RECOMMENDED"},
    {"label": "bug", "confidence": "RECOMMENDED"},
    {"label": "backend", "confidence": "RECOMMENDED"},
    {"label": "database", "confidence": "RECOMMENDED"}
  ],
  "suggested_priority": {
    "priority": "critical",
    "confidence": "RECOMMENDED",
    "rationale": "OWASP Top 10 vulnerability, potential data breach"
  },
  "acceptance_criteria": [
    {"text": "All search inputs use parameterized queries"},
    {"text": "Security audit confirms no SQL injection vectors"},
    {"text": "Existing data integrity validated after fix"}
  ],
  "estimated_complexity": {
    "level": "low",
    "confidence": "RECOMMENDED",
    "rationale": "Standard fix: use query parameters instead of string concatenation"
  }
}

Integration Points

PilotSpaceAgent: Orchestrator routes to this skill via intent detection or /enhance-issue command
MCP Tools: Uses search_issues to find similar issues for label suggestions
Approval Flow: Enhancement suggestions are AUTO_EXECUTE (non-destructive)
Label Management: Checks workspace custom labels before suggesting

References

Design Decision: DD-048 (Confidence Tagging)
Design Decision: DD-003 (Auto-execute for suggestions)
Schema: backend/src/pilot_space/ai/sdk/output_schemas.py