Enforce policies on MCP tool calls and capture portable evidence of execution. Built for developers who need audit trails, compliance checks, and reproducible records of model interactions.
io.github.Rul1an/assay-mcp-server
Local install
STDIO
No auth required
How models use it and what it is built for.
Enforce policies on MCP tool calls and capture portable evidence of execution. Built for developers who need audit trails, compliance checks, and reproducible records of model interactions.
Local install — runs as a subprocess.
Where to find authoritative docs and source for Assay.
Paste any of these into Agent Studio after connecting Assay.
Common questions about connecting and running Assay.
What is a policy-enforcing MCP proxy?
It's a middleware layer that sits between an MCP client and servers, validating tool calls against configurable policies before they execute. It also generates portable evidence (audit logs) of what ran and why calls were allowed or blocked.
What does 'portable evidence output' mean?
The server captures structured records of each tool invocation—including the call, the policy decision, and execution result—in a format you can export, audit, or replay independently of the original MCP session.
How do I install and run assay-mcp-server?
Download the prebuilt binary from the GitHub releases page (v3.9.2 for Linux is available), then configure it as an MCP server in your client. It communicates via stdio transport.
Can I use this for compliance or audit requirements?
Yes—the portable evidence output is designed for audit trails and compliance workflows. You can log and review all tool calls and policy decisions in a structured, reproducible format.
What policy rule formats does it support?
The registry metadata does not specify the exact policy syntax. Refer to the GitHub repository documentation or source code for details on rule configuration and supported constraints.
MCP Playground runs 10,000+ hosted MCP servers — GitHub, Linear, Notion, Stripe, Sentry and more — across Claude, GPT, Gemini, DeepSeek and 30+ AI models. Compare model answers side-by-side, save agent presets, share runs. Zero install.
Open Agent Studio