StratVerity MCP Server
# StratVerity MCP Server
> Audit and verify algorithmic trading backtests — directly from Claude Code,
> Cursor, Windsurf, and any MCP client.
StratVerity MCP exposes backtest-audit tools to LLM coding agents. It **delegates
all computation to the StratVerity FastAPI backend** (`https://api.stratverity.com`)
and never executes submitted strategy code locally.
## Tools
| Tool | Description |
|---|---|
| `stratverity_quick_check` | Static diagnostic of a strategy excerpt (look-ahead bias, repainting, overfitting). |
| `stratverity_verify_hash` | Verify the authenticity of a sealed audit proof by SHA-256 hash. |
| `stratverity_get_badge` | Generate the embeddable badge URL for a certified audit. |
## Install
### Smithery (recommended)
```
npx -y @stratverity/stratverity-mcp
```
### Claude Code
```json
// .mcp.json
{
"mcpServers": {
"stratverity": {
"command": "npx",
"args": ["-y", "@stratverity/stratverity-mcp"]
}
}
}
```
### Cursor / Windsurf
Add the same server entry to your MCP config, pointing at `npx -y @stratverity/stratverity-mcp`.
## Environment
| Variable | Required | Description |
|---|---|---|
| `STRATVERITY_API_URL` | no | Backend origin (default `https://api.stratverity.com`) |
| `STRATVERITY_API_KEY` | no | Bearer token for authenticated endpoints |
## Development
```bash
npm install
npm run build # compile to dist/
npm run dev # run with tsx
```
## Security
- Fail-closed: any upstream error returns a structured error, never a fabricated result.
- No code execution, no live trading, no secret handling.
- Rate-limited by the upstream backend.
## License
MIT
TDQS
Scored across 3 tools
Each tool targets a clearly distinct function: quick_check performs static diagnostics, verify_hash authenticates proofs, and get_badge generates badge URLs. There is no overlap or ambiguity between them.
All tools share the stratverity_ prefix and snake_case style, but quick_check uses an adjective-noun pattern while verify_hash and get_badge follow verb_noun. This is a minor deviation from an otherwise consistent convention.
Three tools is at the lower end but appropriate for the server's narrow audit-verification purpose. Each tool earns its place and the set is well-scoped without unnecessary bloat.
The core workflow is covered: run a quick diagnostic, verify a sealed proof, and generate a badge. A minor gap is the lack of a way to retrieve audit details or history, but agents can work around this for the main use cases.