SDLC Workflow Skill
This skill provides guidance for orchestrating and debugging ai-sdlc workflow execution.
Overview
The ai-sdlc CLI manages story progression through phases (research → planning → implementation → review). This skill helps run workflows effectively and debug issues when stories get stuck.
Workflow Commands
Basic Execution
# Run workflow for a specific story ai-sdlc run --story S-0001 # Run with verbose output (recommended for debugging) ai-sdlc run --story S-0001 --verbose # Dry run - show what would happen without executing ai-sdlc run --story S-0001 --dry-run # Auto mode - skip confirmations ai-sdlc run --story S-0001 --auto
Checking Status
# List all stories ai-sdlc list # Show story details including current phase ai-sdlc show S-0001 # Filter by label ai-sdlc list --label epic-auth
Common Issues
Story Not Progressing
Symptoms: Story stays in same phase after run completes
Debugging Steps:
- •Check story frontmatter status:
ai-sdlc show <story-id> - •Look for phase gate blockers in story.md
- •Check if acceptance criteria are met
- •Run with
--verboseto see phase execution details
Common Causes:
- •Missing required frontmatter fields
- •Phase prerequisites not met
- •Acceptance criteria validation failing
Recovery Limits Hit
Symptoms: "Max recovery attempts reached" error
What's Happening: The workflow has a safety limit on automatic retries
Resolution:
- •Check verbose output for the actual failure
- •Manually fix the underlying issue
- •Reset story state if needed
- •Re-run with
--verbose
Stage Gate Failures
Symptoms: Phase completes but story doesn't advance
Debugging:
- •Check
story.mdfor phase completion markers - •Verify all checklist items are complete
- •Look for validation errors in output
- •Check frontmatter
statusandphasefields match
Debugging Tips
Always Use Verbose Mode When Debugging
ai-sdlc run --story S-0001 --verbose
Verbose mode shows:
- •Phase transitions
- •Agent spawning
- •File operations
- •Validation checks
Check Story Folder Contents
# See what artifacts exist ls -la .ai-sdlc/stories/S-0001/ # Common files: # - story.md (main story document) # - research.md (research phase output) # - plan.md (planning phase output) # - implementation-log.md (implementation notes)
Verify Frontmatter State
The story frontmatter controls workflow progression:
--- id: S-0001 title: Story Title status: in-progress # draft | in-progress | blocked | done phase: planning # research | planning | implementation | review ---
Reset Story State
If a story is stuck, you may need to manually edit frontmatter:
- •Open
.ai-sdlc/stories/<id>/story.md - •Check
statusandphasefields - •Adjust as needed
- •Re-run workflow
Multi-Agent Workflow
The workflow spawns multiple agents for different phases. Debug agent issues by:
- •Check agent output - Verbose mode shows agent communication
- •Verify environment - Agents inherit parent environment
- •Watch for timeouts - Long-running agents may hit limits
- •Check tool availability - Some agents need specific tools enabled
Quick Reference
| Issue | First Step |
|---|---|
| Story stuck | ai-sdlc show <id> to check state |
| Phase not advancing | Run with --verbose |
| Unexpected errors | Check story.md frontmatter |
| Agent failures | Check verbose output for spawn errors |
| Recovery limit | Fix root cause, then re-run |