
Mcp Croit Ceph
io.github.croitPublished by GitHub account croitv0.2.16Updated Oct 24, 2026
MCP server for Croit Ceph cluster management with dynamic OpenAPI tool generation
Overview
Connects an AI assistant to a Croit Ceph cluster's REST API so it can inspect pools, OSDs, servers, S3 buckets and logs in natural language.
- What you can do
- Discover and call cluster API endpoints, with field selection, filtering and drill-down into large responses
- Manage Ceph services, storage pools, servers and S3 buckets through auto-generated category tools
- Search and check cluster logs through VictoriaLogs integration, including pattern and anomaly detection
- Summarize large datasets and cache repeated requests to keep token use low
- When to use it
- Useful when you administer a Croit-managed Ceph cluster and want an assistant to answer health, capacity and troubleshooting questions or run routine read operations without clicking through the web UI.
- Requirements
- Runs locally as a Docker container (image croit/mcp-croit-ceph) or from Python source. Needs the CROIT_HOST cluster URL and a CROIT_API_TOKEN created in the Croit web interface, plus network access to the cluster. Optional OPENAPI_FILE, USE_INCLUDED_API_SPEC, LOG_LEVEL and MCP_ARGS settings.
Installation
In SourceWeft
- Open Mcp Croit Ceph in the dashboard and add it to a workspace.
- Enable the server for the chats that should use its tools.
Desktop only via STDIO. STDIO servers start a local process, so they need the SourceWeft desktop host.
Other MCP clients
Follow the launch instructions in the repository.
README
MCP Croit Ceph
AI-powered management for Croit Ceph clusters via the Model Context Protocol (MCP)
Connect your AI assistant (like Claude) directly to your Croit Ceph cluster for intelligent cluster management, troubleshooting, and monitoring.
What is this?
An MCP server that gives AI assistants access to your Croit Ceph cluster's REST API. Ask your AI to:
- "Show me all pools with errors"
- "List OSDs on server node-5"
- "Search logs for slow requests in the last hour"
- "What's the cluster health status?"
The AI can then interact with your cluster through natural language.
Key Features
🚀 Smart & Efficient
- Automatic token optimization - Responses optimized to save 80-95% on AI token costs
- Field selection - Request only the data you need (e.g., just id + name)
- Built-in filtering - grep-like search without multiple API calls
- Intelligent summaries - Large datasets get smart summaries with drill-down capability
🔍 Advanced Log Search
- Native VictoriaLogs integration for powerful log analysis
- Natural language queries: "Find OSD failures in the last 24 hours"
- Pattern detection and anomaly identification
- Pre-built debug templates for common issues
🛡️ Production Ready
- Role-based access control (ADMIN vs VIEWER)
- Automatic API discovery from your cluster
- Docker support with included OpenAPI spec
- Comprehensive error handling
Quick Start
Using Docker (Recommended)
With Claude Desktop
Add to ~/.config/claude/claude_desktop_config.json:
Basic configuration (fetches OpenAPI spec on startup):
Optimized configuration (recommended - 2s startup instead of 5s):
With debug logging:
Note: Volume mounts require absolute paths (not ~/). To download the OpenAPI spec:
Restart Claude Desktop after changing the configuration.
Configuration
Required Environment Variables
Optional Environment Variables
See .env.example for docker usage template.
Getting an API Token
- Log into your Croit cluster web interface
- Go to Settings → API Tokens
- Create a new token with appropriate permissions
- Copy the token and use it in your configuration
Example AI Conversations
Check cluster health:
Find problems:
Debug issues:
Capacity planning:
Features in Detail
Token Optimization
The server automatically reduces AI token consumption:
How it works:
- Field Selection: Request only needed fields
- Smart Summaries: Large datasets get summaries with drill-down
- Caching: Repeated requests use cached data
- Auto-limiting: Sensible defaults prevent token explosions
Available Tools
The AI has access to these tools:
Core Tools:
list_endpoints- Discover available API endpointscall_endpoint- Make API calls with optimizationsearch_last_result- Drill down into large responses
Category Tools (auto-generated):
manage_services- Ceph servicesmanage_pools- Storage poolsmanage_servers- Cluster serversmanage_s3- S3 buckets- And more...
Log Search:
croit_log_search- Advanced log analysiscroit_log_check- Quick log condition checks
Advanced Usage
Local Development
Using Local OpenAPI Spec
For faster startup or offline development:
Command Line Options
Permissions & Security
The server respects your API token's role:
- ADMIN: Full access to all operations
- VIEWER: Read-only access (no create/delete/update)
- Invalid token: Server exits with error
Admin-only categories: maintenance, servers, config, hooks
Troubleshooting
Token not working:
- Verify token in Croit web interface
- Check token hasn't expired
- Ensure token has correct permissions
Connection issues:
- Verify
CROIT_HOSTis correct (include https://) - Check network connectivity to cluster
- Verify firewall allows connection
No tools showing:
- Check Docker logs for errors
- Verify OpenAPI spec is valid
- Try
--no-permission-checkto test
Enable debug logging:
Documentation
- Architecture - Technical architecture and design
- Token Optimization - How optimization works
- Claude Integration - Tips for using with Claude
Contributing
Contributions welcome! Please:
- Fork the repository
- Create a feature branch
- Make your changes
- Run
blackformatter on Python files - Submit a pull request
License
See LICENSE for details
Support
- Issues: GitHub Issues
- Croit Support: For cluster-specific questions
- MCP Protocol: Model Context Protocol Documentation
Made with ❤️ for the Ceph community
Source: README.md at commit cebfa2e
Tools
0Version history
1- v0.2.16LatestSep 16, 2026
Score
40% of the applicable checks met
How complete this listing is, computed from the registry entry, the endpoint and public sources. Not a rating of the server and not a statement that it is safe or working today: reachability is a dated check, and no installs or reviews go into the score.
- Reached in the last 30 days25 pts
Keep the endpoint answering the MCP handshake; SourceWeft probes it every 30 days.
- Auth declared10 pts
Declare the credential headers in the registry entry, or keep the endpoint answering the probe so its auth can be classified.
- Tools listed15 pts
Answer tools/list without a credential, or ship a package that starts in a sandbox without one.
- DescriptionRequired15 pts
- Logo10 pts
- Documentation15 pts
- License10 pts
Declare a license in the repository or the registry entry.