ChangeFlow is a comprehensive ITIL 4 compliant change management system designed for enterprise IT operations. It provides automated risk assessment, approval workflows, and AI-powered change tracking through the Model Context Protocol (MCP).
- Full change lifecycle management (RFC to implementation)
- CAB (Change Advisory Board) approval workflows
- Risk assessment and impact analysis
- Emergency change procedures
- Standard change catalog
- Audit trail and compliance reporting
- Real-time risk scoring based on change attributes
- Freeze period enforcement
- Dependency tracking
- Rollback planning
- Impact analysis on affected systems
The system exposes 8 ITIL-compliant tools via JSON-RPC 2.0:
create_change_request- Submit new RFCs with auto risk assessmentget_change_request- Query change information by IDlist_change_requests- Dashboard views with filteringassess_risk- Automated risk scoring for proposed changescheck_freeze_period- Verify if date is in change freezeget_approval_status- Check CAB approval progressemergency_override- Break-glass procedures with auditaudit_trail- View change history and compliance logs
Test the MCP server using the Cloudflare AI Playground:
- Visit playground.ai.cloudflare.com
- In the MCP Server URL field, enter:
https://mcp.changeflow.us/v1/sse - Authenticate via GitHub OAuth at
https://mcp.changeflow.us/authorize - The playground will automatically discover all 8 ITIL tools
- Try commands like:
- “Create a change request for database upgrade”
- “Check if December 25th is in a freeze period”
- “Show the audit trail”
For detailed testing with the MCP Inspector:
npx @modelcontextprotocol/inspector@latest
# Connect to: https://mcp.changeflow.us/v1/sse- Authorize: Visit https://mcp.changeflow.us/authorize
- Add to Claude: Settings → Custom Connectors → Add
https://mcp.changeflow.us/v1/sse - Start Using: Access ITIL 4 change management tools directly in Claude
Troubleshooting: See Claude.ai Integration Guide and Sequence Diagrams
# Check system status
curl https://mcp.changeflow.us/health
# List available MCP tools (requires OAuth)
curl -X POST https://mcp.changeflow.us/v1/sse \
-H "Content-Type: application/json" \
-H "Authorization: Bearer YOUR_TOKEN" \
-d '{"jsonrpc":"2.0","method":"tools/list","id":1}'# Clone repository
git clone https://github.com/dsp-dr/guile-changeflow.git
cd guile-changeflow
# Start MCP server (auto-starts on port 8427 with direnv)
node scripts/mcp-local-server.js
# Run test suite
node test/mcp-comprehensive-test.js# Configure Cloudflare credentials
cp .env.example .env
# Edit .env with your CLOUDFLARE_API_TOKEN and CLOUDFLARE_ACCOUNT_ID
# Deploy to Cloudflare Workers
cd infra/cloudflare
wrangler publish{
"jsonrpc": "2.0",
"method": "create_change_request",
"params": {
"title": "Database Schema Update",
"description": "Add authentication fields to user table",
"category": "normal",
"priority": "medium",
"impact": "medium",
"risk_score": 45,
"scheduled_start": "2025-09-20T10:00:00Z",
"scheduled_end": "2025-09-20T12:00:00Z"
},
"id": 1
}{
"jsonrpc": "2.0",
"method": "assess_risk",
"params": {
"change_id": "CHG-20250914-0001",
"factors": ["production", "database", "authentication"]
},
"id": 2
}- Web Dashboard: Executive view with real-time metrics
- MCP Server: AI agent integration layer
- Risk Engine: Automated assessment and scoring
- State Machine: ITIL-compliant workflow engine
- Database: SQLite with full audit trails
submitted → assessing → approved → implementing → completed
↘ rejected ↓ ↗
↘ needs-info → failed ───
↘ cancelled
| Type | Risk Threshold | Approval | Use Case |
|---|---|---|---|
| Standard | < 30 | Automatic | Routine updates |
| Normal | 30-70 | Required | Most changes |
| Emergency | > 70 | Expedited | Critical fixes |
| Factor | Weight | Description |
|---|---|---|
| Production impact | 30 | Changes affecting production systems |
| Data modifications | 25 | Database schema or data changes |
| Security implications | 20 | Authentication, authorization changes |
| Multi-system changes | 15 | Cross-system dependencies |
| Historical success rate | 10 | Past performance of similar changes |
- GNU Guile 3.0+ (for local development)
- Node.js 18+ (for MCP server)
- Cloudflare Workers account (for deployment)
- Make (for build automation)
# Clone and setup
git clone https://github.com/dsp-dr/guile-changeflow.git
cd guile-changeflow
# Environment setup (automatic with direnv)
cp .env.example .env
direnv allow
# Run tests
make test
# Start local MCP server
make mcp-server
# Deploy to Cloudflare
make deploy| Document | Purpose | Location |
|---|---|---|
| Setup & Build | Build system and tooling configuration | docs/setup.org |
| System Design | Architecture and data models | docs/architecture/ |
| ITIL Requirements | ITIL 4 mapping and compliance | docs/itil-requirements.org |
| MCP Protocol | Protocol implementation guide | docs/mcp-protocol.org |
| Deployment | Infrastructure and scaling strategy | docs/deployment.org |
| Roadmap v2.0 | Future development plans | docs/roadmap/V2.0-ROADMAP.org |
- Unit tests: Models and state machine
- Integration tests: MCP protocol
- System tests: End-to-end workflows
- Performance tests: Bulk operations
# All tests
make test
# Specific test suites
guile test/test-core-models.scm
node test/mcp-comprehensive-test.js
./scripts/test-endpoints.sh| Environment | URL | Purpose |
|---|---|---|
| Production | https://mcp.changeflow.us | Live system |
| Staging | https://guile-changeflow-staging.jasonwalsh.workers.dev | Testing env |
| Local | http://localhost:8427 | Development |
- GitHub Actions for testing
- Automatic deployment on main branch
- Release tagging and versioning
See CONTRIBUTING.org for:
- Code style guidelines
- Commit message format
- Pull request process
- Testing requirements
MIT License - See LICENSE for details
- Issues: GitHub Issues
- Documentation: docs/ directory
- API Reference: Online Documentation