Overview
What is Sentinel Signal MCP?
Local stdio MCP server that exposes Sentinel Signal as agent tools: score workflows, check limits and usage, and submit feedback. Designed for AI agents and automated pipelines.
How to use Sentinel Signal MCP?
Recommended install with uvx sentinel-signal-mcp. Set environment variables such as SENTINEL_BASE_URL, SENTINEL_TOKEN_BASE_URL, and optionally SENTINEL_API_KEY (or rely on auto-minted trial key). Run the server and invoke any of the provided tools.
Key features of Sentinel Signal MCP
- Exposes four Sentinel Signal tools as MCP tools
- Supports env-based API key or auto-minted trial key
- Configurable base URLs and timeout
- Credential caching with reset option
- No external runtime dependencies beyond Python
Use cases of Sentinel Signal MCP
- Score structured payloads against named workflows
- Retrieve current plan limits for an API key
- Check usage for a given month
- Submit structured outcome feedback
FAQ from Sentinel Signal MCP
What tools does Sentinel Signal MCP provide?
Four tools: score_workflow, get_limits, get_usage, submit_feedback.
How is authentication handled?
Either set the SENTINEL_API_KEY environment variable, or omit it (and SENTINEL_NO_TRIAL) to auto-mint a trial key at startup.
What do common error codes mean?
402 → trial quota exhausted (upgrade URL returned), 429 → rate limited (retry later), 401/403 → invalid or missing API key (set a key or allow trial minting).
How do I reset cached credentials?
Run sentinel-signal-mcp --reset-credentials to clear saved credentials and obtain a fresh trial key.
What runtime or install command is recommended?
uvx sentinel-signal-mcp is the recommended command for installing and running the server.
