Last Rock MCP
# Last Rock MCP
GARZA OS Last Rock Dev MCP Server - Infrastructure orchestration with state tracking, distributed locking, and automatic rollback.
## Features
š **Infrastructure Orchestration**
- Deploy MCP servers to Fly.io with state tracking
- Deploy Cloudflare Workers with health checks
- Restart services with automatic rollback
- Comprehensive health monitoring
- Auto-recovery playbooks
š **Safety & Reliability**
- Distributed locking prevents concurrent operations
- State tracking in deployments.json
- Pre/post deployment health checks
- Automatic rollback on failure
- Full operation audit trail
š ļø **Tools Included**
1. `deploy_mcp_server` - Deploy to Fly.io with safety checks
2. `deploy_cloudflare_worker` - Deploy Workers with validation
3. `restart_service` - Safe service restarts with rollback
4. `check_services_health` - Monitor service health
5. `trigger_auto_recovery` - Execute recovery playbooks
6. `get_infrastructure_status` - Full infrastructure overview
## Installation
### For Development
```bash
npm install
npm run build
```
### For Production (Fly.io)
```bash
# Deploy to Fly.io
fly deploy
# Or use orchestrator (recommended)
deploy_mcp_server(app_name="lastrock-mcp")
```
## Usage in Claude
These tools are auto-discovered by Claude and preferred over raw `ssh_exec` commands:
```javascript
// ā DON'T DO THIS
ssh_exec("fly deploy --app garza-home-mcp")
// ā
DO THIS INSTEAD
deploy_mcp_server(app_name="garza-home-mcp")
```
## Configuration
The orchestrator operates on:
- **Operations**: `/Users/customer/garza-os-github/operations/`
- **State**: `/Users/customer/garza-os-github/.orchestrator/state.json`
- **Locks**: `/Users/customer/garza-os-github/.orchestrator/locks/`
## Architecture
```
lastrock-mcp/
āāā src/
ā āāā index.ts # MCP server implementation
āāā build/ # Compiled JavaScript
āāā package.json
āāā tsconfig.json
āāā README.md
```
## Testing
```bash
# Health check (safe, read-only)
check_services_health(service_group="all")
# Get infrastructure status
get_infrastructure_status()
# Deploy test (creates real changes!)
deploy_mcp_server(app_name="test-mcp")
```
## Integration
This MCP server integrates with GARZA OS orchestrator:
- Orchestrator: `/Users/customer/garza-os-github/orchestrator/`
- Operations: `/Users/customer/garza-os-github/operations/`
- Documentation: `/Users/customer/garza-os-github/docs/`
## Why This Matters
**Before:**
- Raw SSH commands
- No state tracking
- No safety checks
- No rollback capability
- Manual recovery
**After:**
- Managed orchestration
- Full state tracking
- Automatic health checks
- Automatic rollback
- Auto-recovery playbooks
## Related
- [Infrastructure Orchestrator](https://github.com/itsablabla/garza-os-github)
- [Garza Home MCP](https://github.com/itsablabla/garza-home-mcp)
- [CF Workers MCP](https://github.com/itsablabla/cf-workers-mcp)
## License
MIT - Jaden Garza
TDQS
Scored across 6 tools
Each tool has a clearly distinct purpose with no overlap: health checking, two specific deployment tools (Cloudflare Worker and MCP server), infrastructure status overview, service restart, and auto-recovery. The descriptions reinforce distinct domains (deployment vs. operations vs. monitoring).
Tools follow a consistent verb_noun pattern (e.g., check_services_health, deploy_cloudflare_worker) with clear action-object naming. The only minor deviation is 'get_infrastructure_status' which uses 'get' instead of 'check' like the health tool, but the pattern remains readable and predictable.
Six tools is well-scoped for infrastructure management, covering deployment, monitoring, and recovery workflows. Each tool earns its place with no redundancy, and the count aligns with typical server ranges (3-15 tools) for this domain.
The toolset provides strong coverage for infrastructure lifecycle: deployment (two tools), health monitoring, status overview, restart, and recovery. A minor gap exists in lacking update/delete operations for deployed resources, but core workflows are well-covered with safe alternatives to raw SSH commands.