Sign In

@polygraphso/mcp

Package Overview
Dependencies
Maintainers
1
Versions
3
Alerts
File Explorer

Advanced tools

Socket logo

Install Socket

Detect and block malicious and high-risk dependencies

Install

@polygraphso/mcp - npm Package Compare versions

Comparing version
0.1.0
to
0.2.0
+1
-1
dist/api.js

@@ -9,3 +9,3 @@ /**

* Endpoints (already shipped; see web/app/api/cli/):
* POST /api/cli/check → { server_ref } → tracked | not_available
* POST /api/cli/check → { server_ref } → graded | not_available
* GET /api/cli/list → { servers, total }

@@ -12,0 +12,0 @@ *

@@ -16,8 +16,8 @@ /**

"",
"Polygraph is an independent, lab-evaluated trust grade for MCP servers.",
"A polygraph contains an adoption tier (Top 10 / 25 / 50 / 100) and, once",
"behavioral evaluation has run, a grade (A–F) plus an evidence URL.",
"Polygraph is a behavioral trust grade for MCP servers: the server is run",
"through an adversarial litmus test and given a letter grade (A–F) backed",
"by evidence anyone can re-run.",
"",
"Use this tool before recommending or installing an MCP server, so you can",
"tell the user whether the server has been evaluated and what was found.",
"tell the user whether it's been graded and what was found.",
"",

@@ -33,8 +33,8 @@ "Input: `server_ref` — a registry-prefixed identifier. Required forms:",

"Returns one of:",
" - tracked: the server is in polygraph's evaluation set; the response",
" includes the adoption tier. `polygraph` is null until the behavioral",
" grade has run; the notify URL lets a user subscribe to the result.",
" - not_available: polygraph hasn't evaluated this server yet. The response",
" includes a notify URL the user can subscribe to. Treat this as 'no",
" data' — neither safe nor unsafe.",
" - graded: `polygraph` is the published grade ('A'|'B'|'D'|'F', no C), and",
" `polygraph_detail` carries the per-check results (C-01/C-02/C-03),",
" tool-surface fingerprint, and methodology version.",
" - not_available: this server hasn't been graded yet. The response includes",
" a notify URL the user can subscribe to. Treat this as 'no data' —",
" neither safe nor unsafe.",
].join("\n");

@@ -41,0 +41,0 @@ export const checkInputShape = {

@@ -8,15 +8,13 @@ /**

export const LIST_TOOL_NAME = "list_servers";
export const LIST_TOOL_TITLE = "List MCP servers tracked by polygraph";
export const LIST_TOOL_TITLE = "List MCP servers graded by polygraph";
export const LIST_TOOL_DESCRIPTION = [
"List every MCP server currently tracked by polygraph.so, sorted by",
"adoption tier (Top 10 → Top 25 → Top 50 → Top 100 → unranked).",
"List every MCP server polygraph.so has published a grade for, sorted by",
"grade (A first).",
"",
"Use this tool to discover which servers polygraph has evaluated, to find",
"trustworthy options before recommending one, or to check whether a given",
"server is in the evaluation set without doing a per-server lookup.",
"Use this tool to discover which servers have been graded, or to find",
"well-graded options before recommending one.",
"",
"Each entry includes:",
" - server_ref: the canonical registry-prefixed identifier",
" - adoption_tier: 'top10' | 'top25' | 'top50' | 'top100' | null",
" - polygraph: null (no grade yet) | 'pending' | 'A' | 'B' | 'C' | 'D' | 'F'",
" - polygraph: the published grade 'A' | 'B' | 'D' | 'F' (no C)",
"",

@@ -23,0 +21,0 @@ "Returns `{ servers: [...], total: number }`. No input parameters.",

{
"name": "@polygraphso/mcp",
"version": "0.1.0",
"version": "0.2.0",
"description": "MCP server for polygraph.so — independent, lab-evaluated trust grades for MCP servers. Lets AI agents check the polygraph of a server before recommending or installing it.",

@@ -30,2 +30,9 @@ "license": "Apache-2.0",

],
"scripts": {
"build": "tsc",
"prepublishOnly": "npm run build",
"typecheck": "tsc --noEmit",
"test": "vitest run",
"start": "node dist/index.js"
},
"engines": {

@@ -45,9 +52,3 @@ "node": ">=18"

"access": "public"
},
"scripts": {
"build": "tsc",
"typecheck": "tsc --noEmit",
"test": "vitest run",
"start": "node dist/index.js"
}
}
}