Skip to main content
Glama
polygraphso

Polygraph

Official
by polygraphso
OverviewSchemaRelated ServersScoreDiscussions
TypeScript

Run a behavioral litmus on an MCP server

run_litmus

Grade any MCP server A–F by running four security checks: tool-output injection, permission overreach, sensitive-data handling, and adversarial-input resilience. Uses Docker sandboxing when available.

Instructions

Grade an MCP server A–F against the open behavioral litmus (litmus-v5). The harness connects the way an agent would, fingerprints the tool surface, and runs four checks: C-01 tool-output injection, C-02 permission/egress overreach (egress in a hardened default-deny Docker sandbox, plus a declared-permission honesty check), C-03 sensitive-data handling (planted canaries), and C-04 adversarial-input handling (malformed/oversized and jailbreak inputs).

This is ACTIVE: it launches the target server's code to exercise it (egress- sandboxed when Docker is available) and takes ~20–60s. It is not a lookup — for a server's already-published grade, use verify_attestation. No wallet or RPC needed.

server_ref examples: npm/@modelcontextprotocol/server-filesystem · https://example.com/mcp · ./build/index.js. For a token-gated https:// target, pass bearer. If Docker is unavailable, C-02 is skipped and the grade is capped at B for that run.

Input Schema

TableJSON Schema
NameRequiredDescriptionDefault
server_refYesWhat to grade: a registry ref (npm/@scope/server), an https:// MCP URL, or a local path to an MCP entry file.
bearerNoBearer token for a token-gated https:// MCP server. Sent as `Authorization: Bearer <token>` to the target origin only. Ignored for stdio/local targets.
headerNoExtra HTTP headers for a gated https:// target, each "Key: Value" (e.g. "X-Api-Key: …"). Overrides the bearer-derived Authorization for the same key. Ignored for stdio/local targets.
unsafe_host_execNoRequired to grade a registry ref or local path: it launches the target's own code, and without Docker isolation that runs on THIS host. Set true to accept host execution. Ignored for https:// targets or when LITMUS_STDIO_ISOLATION=docker.
timeout_secondsNoAggregate wall-clock ceiling for the whole run, in seconds (default 900). Bounds a hostile server that stretches the run across many tools/probes.

Tool Definition Quality

A4.6/5.0
Behavior5/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

Annotations provide openWorldHint=true but little else. The description excellently fills in behavioral details: it is active (launches code), takes 20-60s, uses Docker sandbox for egress isolation, performs four specific checks, and caps grade if Docker unavailable. This is comprehensive and exceeds the annotation burden.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness4/5

Is the description appropriately sized, front-loaded, and free of redundancy?

The description is well-structured, starting with core purpose, then listing checks, behavioral notes, and examples. It is front-loaded but slightly long; each sentence earns its place, though minor redundancy could be trimmed.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness5/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

Given the tool's complexity (active, multi-check, safety constraints), the description is thorough. It explains what the tool does, how it works, safety implications, output (grade), and limitations. No output schema exists, but description adequately covers return value scope.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters4/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

Schema covers all 5 parameters (100% coverage), baseline is 3. The description adds value by providing examples for server_ref, clarifying bearer token usage, explaining header override, highlighting unsaf e_host_exec requirements, and specifying default timeout_seconds. This extra context improves parameter understanding.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose5/5

Does the description clearly state what the tool does and how it differs from similar tools?

The description clearly states the tool's purpose: grading an MCP server A-F against a specific litmus version. It uses a specific verb ('grade') and resource ('MCP server'), and distinguishes from sibling tools like verify_attestation for published grades. This leaves no ambiguity.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines4/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

The description explains when to use this tool (active grading) and explicitly contrasts with verify_attestation for lookup. It provides guidance on parameters (server_ref examples, bearer token usage, Docker dependency) and mentions sibling tools run_skill_litmus and verify_skill_attestation, though it could more explicitly state when to prefer those.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.

Install Server

Other Tools

Latest Blog Posts

MCP directory API

We provide all the information about MCP servers via our MCP API.

curl -X GET 'https://glama.ai/api/mcp/v1/servers/polygraphso/litmus'

If you have feedback or need assistance with the MCP directory API, please join our Discord server