Cimon-MCP
# Cimon-MCP
MCP server with offline runbooks for high-volume logs, metrics, traces, and event streams. No database or API keys.
## Tools
| Tool | What it does |
|------|----------------|
| `search_solutions` | Search playbooks by keyword |
| `get_solution` | Full entry by id (aliases like `write-amplification` work) |
| `diagnose_symptoms` | Match symptom text to playbooks |
| `health_check` | Entry count and server status |
Resources: `cimon://playbook/{id}` (JSON per entry).
**Topics:** ingestion batching, write amplification, slow queries, schema/sort keys, low-cardinality columns, observability layout, replication lag, query memory limits.
## Setup
Node 18+.
```bash
npm install # runs build via prepare
npm start
```
```bash
npm run dev # rebuild + start
npm test # unit tests + MCP smoke
npm run smoke # MCP integration only
```
## Cursor
`.cursor/mcp.json` or Settings → MCP. Point `args` at your `dist/index.js`:
```json
{
"mcpServers": {
"Cimon-MCP": {
"command": "node",
"args": ["/absolute/path/to/Cimon-MCP/dist/index.js"]
}
}
}
```
See `mcp.json.example`.
## License
MIT
TDQS
Scored across 4 tools
The tools are mostly distinct: search_solutions and get_solution are clearly search vs. fetch, while health_check is administrative. diagnose_symptoms overlaps somewhat with search_solutions as both return playbook recommendations, but the descriptions clarify the different input modes (keyword search vs. symptom mapping).
All tool names follow a consistent snake_case verb_noun pattern: search_solutions, get_solution, diagnose_symptoms, health_check. No mixed conventions or vague verbs.
Four tools is well-scoped for a focused playbook server. Each tool has a clear purpose and the count feels neither sparse nor bloated.
The core read-only workflows are covered: search, fetch by id, and symptom-based diagnosis. A minor gap is the absence of a list_all_playbooks tool, but search_solutions likely covers exploration. No create/update/delete is expected for this offline reference server.