Pit Crew SI - Demo Script

This script guides you through a complete demonstration of Pit Crew Supreme Intelligence.

Setup (5 minutes)

Step 1: Deploy the App

# From the project directory
npm install
forge deploy
forge install

Step 2: Access the Dashboard

1. Navigate to your Jira instance
2. Click "Apps" → "Pit Crew SI Dashboard"
3. Verify the dashboard loads with initial statistics

Step 3: Verify Configuration

The dashboard should show:

• Mode: Dry-Run (safe default)
• Auto-Approve: Disabled
• Total Actions: 0

Demo 1: Automatic Bug Triage (5 minutes)

Scenario

A critical bug is reported that needs immediate attention.

Steps

1. Create a bug in Jira:

   • Title: "Critical: Authentication API returning 500 errors"
   • Description: "Users unable to login. API endpoint /auth/login failing with 500 internal server error. Urgent fix needed for production."
   • Type: Bug
   • Priority: (Leave as default, let Pit Crew SI adjust it)

2. Watch the magic happen:

   • Within seconds, Pit Crew SI detects the new issue
   • Automatically analyzes the content
   • Identifies keywords: "critical", "api", "authentication", "urgent", "production"

3. Check the Dashboard:

   • Refresh the Pit Crew SI Dashboard
   • You should see new action logs:
     • "Label Issue" - Proposes adding: bug, api, security, critical-bug
     • "Set Priority" - Proposes changing to "High" or "Highest"

4. Review the Proposal:

   • Click on the action to see the explanation
   • Note the reasoning: "Issue contains critical keywords and production impact"
   • Observe that it's in "proposed" status (dry-run mode)

Expected Results

• ✅ Issue automatically triaged
• ✅ Priority identified as critical/high
• ✅ Relevant labels suggested
• ✅ Actions logged with explanations
• ✅ Guardrails checked and documented

Key Points to Highlight

• Intelligent Analysis: Understands context from keywords
• Explainable AI: Every decision has clear reasoning
• Safety First: Dry-run mode prevents unwanted changes
• Full Auditability: Complete log of what was analyzed and proposed

Demo 2: Sprint Planning (7 minutes)

Scenario

Planning a 2-week sprint with mixed priorities.

Preparation

Create 10-15 issues in Jira with varying priorities:

High Priority:

1. "Fix memory leak in user service"
2. "Critical security vulnerability in auth module"
3. "Database migration for new schema"

Medium Priority: 4. "Implement user profile page" 5. "Add pagination to search results" 6. "Optimize database queries for reports" 7. "Update API documentation"

Low Priority: 8. "Refactor legacy code in utils module" 9. "Add unit tests for helper functions" 10. "Improve error messages"

Steps

1. Collect Issues:

   // Use Jira API or manually note issue keys
   const issues = [
     'PROJ-101', 'PROJ-102', 'PROJ-103',
     // ... etc
   ];
   

2. Generate Sprint Plan:

   const plan = await invoke('generateSprintPlan', {
     issues: issueObjects, // Full issue objects
     goals: [
       'Stabilize production environment',
       'Complete security improvements',
       'Improve user experience'
     ],
     capacity: 40 // Story points
   });
   

3. Review the Plan:

   • Open the sprint plan results
   • Observe how issues are prioritized
   • Check capacity utilization
   • Review identified risks

Expected Results

• ✅ Issues ranked by priority score
• ✅ Capacity respected (won't overcommit)
• ✅ Risks identified (e.g., "too many bugs indicates technical debt")
• ✅ Dependencies noted
• ✅ Clear reasoning for each selection

Key Points to Highlight

• Data-Driven: Uses priority, type, age, and content
• Capacity-Aware: Respects team limits
• Risk Assessment: Identifies potential issues
• Goal-Aligned: Considers sprint objectives

Demo 3: Execution Brief Generation (5 minutes)

Scenario

Breaking down a complex feature into actionable tasks.

Steps

1. Create a Feature Issue:

   • Title: "Implement OAuth2 authentication"
   • Description: "Add OAuth2 support for Google and GitHub login. Users should be able to sign in using their existing accounts. Include proper error handling and security measures."

2. Generate Execution Brief:

   const brief = await invoke('generateExecutionBrief', {
     issue: featureIssue,
     relatedPages: [] // Or include relevant Confluence pages
   });
   

3. Review the Brief:

   • Title and summary
   • Clear objectives
   • Broken-down tasks with estimates
   • Technical context
   • References to related resources

Expected Results

Execution Brief: Implement OAuth2 authentication

Objectives:
- Complete OAuth2 authentication implementation
- Ensure code quality and test coverage
- Update relevant documentation

Tasks:
1. Analyze requirements and design approach (2 hours)
2. Implement core functionality (1 day)
3. Write tests (4 hours)
4. Code review and refinement (2 hours)

Technical Context:
- Issue Type: Story
- Project: PROJ
- Status: To Do

Key Points to Highlight

• Structured Approach: Clear task breakdown
• Time Estimates: Helps with planning
• Context Preservation: Includes all relevant info
• Actionable: Ready to execute

Demo 4: Safe Action Execution (8 minutes)

Scenario

Executing approved actions with full safety checks.

Steps

1. Switch to Execute Mode:

   • In the Dashboard, click "Toggle Mode"
   • Confirm the switch to Execute mode
   • Observe the mode badge changes to green

2. Create a New Bug:

   • Title: "Button styling broken on mobile devices"
   • Description: "The submit button on the login form appears cut off on mobile screens. Needs UI fix."

3. Watch Auto-Triage:

   • Pit Crew SI analyzes the issue
   • Proposes actions: Add labels bug, ui, mobile
   • Set priority to Medium

4. Review Guardrails:

   • Check the action log
   • See guardrail checks:
     • ✅ Action Allowed: label-issue is permitted
     • ✅ Label Count: 3 labels (within limit)
     • ✅ Reserved Labels: No reserved labels detected
     • ✅ Impact Assessment: Low impact

5. Approve and Execute:

   • If auto-approve is disabled, manually approve
   • Action executes
   • Labels are added to the Jira issue

6. Verify in Jira:

   • Open the issue in Jira
   • Confirm labels were added: bug, ui, mobile
   • Check action log shows "executed" status

7. Demonstrate Revert:

   • Click "Revert" on the action
   • Labels are removed
   • New "reverted" log entry created

Expected Results

• ✅ Actions execute in execute mode
• ✅ Guardrails prevent unsafe actions
• ✅ Changes appear in Jira immediately
• ✅ Full audit trail maintained
• ✅ Revert functionality works

Key Points to Highlight

• Dual Mode Operation: Safe testing with dry-run
• Guardrails: Multiple safety checks
• Real Changes: Actually modifies Jira
• Reversible: Can undo actions
• Audit Trail: Complete history

Demo 5: Guardrails in Action (5 minutes)

Scenario

Demonstrating that guardrails prevent unsafe actions.

Steps

1. Create High-Priority Override:

   • Create a bug with "minor typo in footer"
   • Pit Crew SI suggests low priority
   • Manually try to create an action setting priority to "Blocker"

2. Run Guardrails:

   const proposal = {
     action: 'set-priority',
     details: {
       targetKey: 'PROJ-999',
       changes: { priority: 'Blocker' }
     },
     // ...
   };
   
   const guardrails = await GuardrailsService.checkAction(proposal, config);
   

3. Review Results:

   • Guardrail fails: "Setting priority to Highest/Blocker requires manual approval"
   • Action status: "proposed" (not executed)
   • Requires approval even in execute mode

4. Try Reserved Label:

   • Create proposal to add security label
   • Guardrail detects reserved label
   • Blocks automatic execution

Expected Results

• ❌ High-priority actions blocked
• ❌ Reserved labels require approval
• ✅ System stays safe
• ✅ Clear explanations provided

Key Points to Highlight

• Safety First: Prevents mistakes
• Smart Guardrails: Context-aware checks
• Approval Workflow: For sensitive actions
• Configurable: Can adjust rules

Wrap-Up: Key Takeaways (3 minutes)

What We Demonstrated

1. Intelligent Triage

   • Automatic issue analysis
   • Context-aware priority setting
   • Smart label suggestions

2. Sprint Planning

   • Data-driven prioritization
   • Capacity management
   • Risk identification

3. Execution Briefs

   • Task breakdown
   • Time estimation
   • Structured approach

4. Safe Action Execution

   • Dual mode (dry-run/execute)
   • Guardrails system
   • Full auditability
   • Reversible actions

5. Developer-First Design

   • Explainable AI
   • Complete transparency
   • Easy configuration
   • Forge-native integration

Value Proposition

For Teams:

• ⏱️ Saves Time: Automatic triage and planning
• 🎯 Improves Focus: Surfaces important work
• 📊 Better Planning: Data-driven sprint plans
• 🔒 Maintains Safety: Guardrails prevent mistakes

For Individuals:

• 🤖 AI Assistant: Always-on intelligence
• 📝 Documentation: Auto-generated briefs
• 🔍 Context: Understands your workflow
• ✅ Actionable: Turns insights into actions

Next Steps

1. Customize: Adjust triage logic for your team
2. Integrate: Connect with Rovo Agents
3. Extend: Add custom actions
4. Scale: Deploy across projects

---

Questions to Ask the Audience:

• What manual triage tasks take the most time for your team?
• What actions would you want the AI to perform automatically?
• How would you customize the guardrails for your workflow?

Resources:

• README.md - Complete documentation
• API.md - API reference
• SETUP.md - Setup guide
• Dashboard - Live monitoring and control