Fetch GitHub Issue Analysis

Purpose

Retrieve complete GitHub issue information including title, body, labels, and AI-generated analysis comment (if present) for use in conductor workflow Phase 1.

When to Use

• Conductor workflow Phase 1 (Issue Discovery and Planning)
• Workflow resumption with existing issue number
• Issue selection and validation
• Architecture planning with AI insights

Instructions

Step 1: Fetch Basic Issue Data

# Safe parameter handling with validation
ISSUE_NUMBER=${1:-""}

# Validate required parameter
if [ -z "$ISSUE_NUMBER" ]; then
  echo "❌ Error: Issue number required"
  echo "Usage: fetch-github-issue-analysis <issue_number>"
  exit 1
fi

# Get issue details
ISSUE_DATA=$(gh issue view "$ISSUE_NUMBER" --json title,body,labels,number,state)

# Extract fields
ISSUE_TITLE=$(echo "$ISSUE_DATA" | jq -r '.title')
ISSUE_BODY=$(echo "$ISSUE_DATA" | jq -r '.body')
ISSUE_STATE=$(echo "$ISSUE_DATA" | jq -r '.state')
LABELS=$(echo "$ISSUE_DATA" | jq -r '[.labels[].name] | join(",")')

Step 2: Check for AI Analysis Label

# Check if issue has ai-analyzed label
if echo "$LABELS" | grep -q "ai-analyzed"; then
  HAS_AI_ANALYSIS=true
else
  HAS_AI_ANALYSIS=false
fi

Step 3: Fetch AI Analysis Comment (if exists)

if [ "$HAS_AI_ANALYSIS" = true ]; then
  # Get repository name with owner
  REPO_NAME=$(gh repo view --json nameWithOwner --jq .nameWithOwner)

  # Fetch AI analysis comment from github-actions bot
  AI_ANALYSIS=$(gh api "repos/$REPO_NAME/issues/$ISSUE_NUMBER/comments" \
    --jq '.[] | select(.user.login == "github-actions[bot]" and (.body | contains("AI Issue Analysis"))) | .body')

  if [ -n "$AI_ANALYSIS" ]; then
    AI_ANALYSIS_FOUND=true
  else
    AI_ANALYSIS_FOUND=false
  fi
else
  AI_ANALYSIS=""
  AI_ANALYSIS_FOUND=false
fi

Step 4: Return Structured Output

Output as JSON for easy parsing by conductor:

{
  "issue": {
    "number": 137,
    "title": "Add user dark mode preference toggle",
    "body": "User wants to...",
    "state": "open",
    "labels": ["feature", "frontend", "ai-analyzed"]
  },
  "aiAnalysis": {
    "found": true,
    "content": "# AI Issue Analysis\n\n## Architectural Alignment...",
    "sections": {
      "architecturalAlignment": "...",
      "technicalFeasibility": "...",
      "implementationSuggestions": "...",
      "filesAffected": [...],
      "testingStrategy": "..."
    }
  }
}

Step 5: Parse AI Analysis Sections (Optional)

If AI analysis found, extract key sections:

# Extract Architectural Alignment section
ARCH_ALIGNMENT=$(echo "$AI_ANALYSIS" | sed -n '/## Architectural Alignment/,/## /p' | sed '$d')

# Extract Technical Feasibility section
TECH_FEASIBILITY=$(echo "$AI_ANALYSIS" | sed -n '/## Technical Feasibility/,/## /p' | sed '$d')

# Extract Implementation Suggestions section
IMPL_SUGGESTIONS=$(echo "$AI_ANALYSIS" | sed -n '/## Implementation Suggestions/,/## /p' | sed '$d')

# Extract Files/Components section
FILES_AFFECTED=$(echo "$AI_ANALYSIS" | sed -n '/## Files/,/## /p' | sed '$d')

# Extract Testing Strategy section
TESTING_STRATEGY=$(echo "$AI_ANALYSIS" | sed -n '/## Testing Strategy/,/## /p' | sed '$d')

Output Format

Success Case

{
  "status": "success",
  "issue": {
    "number": 137,
    "title": "Add user dark mode preference toggle",
    "state": "open",
    "labels": ["feature", "frontend", "ai-analyzed"]
  },
  "aiAnalysis": {
    "found": true,
    "architecturalAlignment": "Aligns with component-based architecture...",
    "technicalFeasibility": "High - uses existing React patterns...",
    "implementationSuggestions": "1. Add darkMode to state\n2. Create toggle component...",
    "filesAffected": ["src/components/Settings.tsx", "src/context/WorldContext.tsx"],
    "testingStrategy": "Unit tests for toggle, integration tests for persistence"
  }
}

No AI Analysis Case

{
  "status": "success",
  "issue": {
    "number": 123,
    "title": "Fix character portrait loading",
    "state": "open",
    "labels": ["bug", "frontend"]
  },
  "aiAnalysis": {
    "found": false
  }
}

Integration with Conductor

Used in conductor Phase 1, Step 1:

**Step 1: Issue Selection**

If issue number provided by user:

Use `fetch-github-issue-analysis` skill:
- Input: issue_number
- Output: issue details + AI analysis (if available)

If AI analysis found:
  ✅ Use AI insights for architecture planning
  📋 Extract: architectural alignment, implementation suggestions, testing strategy
  → Skip orchestrator selection, proceed to architecture review

If no AI analysis:
  → Delegate to orchestrator for issue selection logic

Error Handling

Issue Not Found

{
  "status": "error",
  "error": "Issue #999 not found",
  "code": "NOT_FOUND"
}

Issue Closed

{
  "status": "warning",
  "issue": {...},
  "warning": "Issue is closed - confirm before proceeding"
}

Rate Limit

{
  "status": "error",
  "error": "GitHub API rate limit exceeded",
  "code": "RATE_LIMIT",
  "retryAfter": 3600
}

Related Skills

• parse-ai-analysis

  - Deep parsing of AI analysis sections

• select-optimal-issue

  - Backlog selection when no issue specified

• create-tracking-issue

  - Create new issues

Examples

Example 1: Issue with AI Analysis

# Input
fetch-github-issue-analysis 137

# Output
{
  "status": "success",
  "issue": {
    "number": 137,
    "title": "Add user dark mode preference toggle"
  },
  "aiAnalysis": {
    "found": true,
    "architecturalAlignment": "Aligns with React Context patterns..."
  }
}

Example 2: Issue without AI Analysis

# Input
fetch-github-issue-analysis 123

# Output
{
  "status": "success",
  "issue": {
    "number": 123,
    "title": "Fix character portrait loading"
  },
  "aiAnalysis": {
    "found": false
  }
}

Best Practices

1. Always check issue state - Warn if closed
2. Cache AI analysis - Avoid repeated API calls
3. Handle rate limits - Implement exponential backoff
4. Validate JSON output - Ensure well-formed response
5. Extract sections carefully - AI analysis format may vary

Notes

• AI analysis format is defined by github-actions bot
• Comment must contain "AI Issue Analysis" heading
• Sections may vary based on issue type
• Always validate AI analysis exists before parsing sections