Mcp Croit Ceph — independently scanned and version-tracked by SaferSkills.
SaferSkills independently audited Mcp Croit Ceph (Agent Skill) and scored it 45/100 (orange). The audit ran 55 deterministic rules across Security, Supply Chain, Maintenance, Transparency, and Community; it found 1 high-severity and 0 lower-severity findings. The full rule-by-rule trace and per-finding evidence are below. Free, methodology-open.
Findings & checks · 1 flagged
A base64 string of 128+ characters appears in a documentation file. Encoded prompt injection hides the hostile instruction in base64 — invisible to keyword filters — and relies on the agent's ability to decode it at runtime. There is no normal authoring reason to embed a multi-hundred-byte base64 blob in skill docs.
*.sig, SIGNATURES) outside the documentation.Every scanned point with the score it earned and what moved between them.
First recorded scan — no prior version to compare against.
The primary manifest — the file an agent reads to learn what this artifact does.
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.
An MCP server that gives AI assistants access to your Croit Ceph cluster's REST API. Ask your AI to:
The AI can then interact with your cluster through natural language.
docker run --rm -i \
-e CROIT_HOST="https://your-cluster.com" \
-e CROIT_API_TOKEN="your-api-token" \
croit/mcp-croit-ceph:latestAdd to ~/.config/claude/claude_desktop_config.json:
Basic configuration (fetches OpenAPI spec on startup):
{
"mcpServers": {
"croit-ceph": {
"command": "docker",
"args": [
"run", "--rm", "-i",
"-e", "CROIT_HOST=http://your-cluster.croit.io:8080",
"-e", "CROIT_API_TOKEN=your-api-token-here",
"croit/mcp-croit-ceph:latest"
]
}
}
}Optimized configuration (recommended - 2s startup instead of 5s):
{
"mcpServers": {
"croit-ceph": {
"command": "docker",
"args": [
"run", "--rm", "-i",
"-v", "/absolute/path/to/openapi.json:/config/openapi.json:ro",
"-e", "CROIT_HOST=http://your-cluster.croit.io:8080",
"-e", "CROIT_API_TOKEN=your-api-token-here",
"-e", "OPENAPI_FILE=/config/openapi.json",
"croit/mcp-croit-ceph:latest"
]
}
}
}With debug logging:
{
"mcpServers": {
"croit-ceph": {
"command": "docker",
"args": [
"run", "--rm", "-i",
"-e", "CROIT_HOST=http://your-cluster.croit.io:8080",
"-e", "CROIT_API_TOKEN=your-api-token-here",
"-e", "LOG_LEVEL=DEBUG",
"croit/mcp-croit-ceph:latest"
]
}
}
}Note: Volume mounts require absolute paths (not ~/). To download the OpenAPI spec:
curl -H "Authorization: Bearer YOUR_TOKEN" \
http://your-cluster.croit.io:8080/api/swagger.json > openapi.jsonRestart Claude Desktop after changing the configuration.
CROIT_HOST="https://your-cluster.com" # Your Croit cluster URL
CROIT_API_TOKEN="your-api-token" # API token from Croit# Use local OpenAPI spec for faster startup (2s vs 5s)
OPENAPI_FILE="/config/openapi.json"
# Use bundled spec (no external fetch, for offline/dev)
USE_INCLUDED_API_SPEC="true"
# Logging verbosity: DEBUG, INFO, WARNING, ERROR
LOG_LEVEL="INFO"
# Pass additional CLI arguments
MCP_ARGS="--no-permission-check --max-category-tools 5"See .env.example for docker usage template.
Check cluster health:
You: "What's the current cluster status?"
AI: Calls list_endpoints → Finds status endpoint → Returns health summaryFind problems:
You: "Show me all pools with errors"
AI: Calls /pools with fields=["id","name","status"] and filter for errors
Returns only the 3 pools with issues (instead of all 100 pools)Debug issues:
You: "Search logs for OSD failures in the last hour"
AI: Uses croit_log_search with smart query parsing
Returns summary + drill-down capability for detailsCapacity planning:
You: "Which pools are over 80% full?"
AI: Gets pools → Filters by usage → Returns list with recommendationsThe server automatically reduces AI token consumption:
| Without Optimization | With Optimization | Savings |
|---|---|---|
| 100 pools, all fields → 4,000 tokens | 100 pools, id+name → 300 tokens | 92% |
| 500 OSDs, full data → 15,000 tokens | Smart summary → 1,000 tokens | 93% |
How it works:
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 responsesCategory Tools (auto-generated):
manage_services - Ceph servicesmanage_pools - Storage poolsmanage_servers - Cluster serversmanage_s3 - S3 bucketsLog Search:
croit_log_search - Advanced log analysiscroit_log_check - Quick log condition checks# Clone repository
git clone https://github.com/croit/mcp-croit-ceph.git
cd mcp-croit-ceph
# Create virtual environment
python3 -m venv venv
source venv/bin/activate
# Install dependencies
pip install -r requirements.txt
# Run directly
export CROIT_HOST="https://your-cluster.com"
export CROIT_API_TOKEN="your-token"
python mcp-croit-ceph.pyFor faster startup or offline development:
# Download spec from your cluster
curl -H "Authorization: Bearer $CROIT_API_TOKEN" \
https://your-cluster/api/swagger.json > openapi.json
# Use local spec
python mcp-croit-ceph.py --openapi-file openapi.jsonpython mcp-croit-ceph.py \
--openapi-file openapi.json \ # Use local OpenAPI spec
--no-permission-check \ # Skip role check (faster)
--max-category-tools 5 # Limit category toolsThe server respects your API token's role:
Admin-only categories: maintenance, servers, config, hooks
Token not working:
Connection issues:
CROIT_HOST is correct (include https://)No tools showing:
--no-permission-check to testEnable debug logging:
export LOG_LEVEL=DEBUG
python mcp-croit-ceph.pyContributions welcome! Please:
black formatter on Python filesSee LICENSE for details
Made with ❤️ for the Ceph community
~30 seconds. Free. No account. Every finding cites a rule and a line of evidence.