MCP App Store
by KeenEthics
OpenGlass icon

OpenGlass MCP Server for Claude and ChatGPT

By Federico Riveroll

Overview

OpenGlass brings transparency to agentic interactions with four public, read-only tools. No OpenGlass account is required. Reveal hidden content in pasted HTML, plain text or a public URL: invisible characters, inline-hidden text, comments, metadata and other matched patterns. URL checks cover the bytes fetched at check time. Scripts do not run; stylesheets, rendered layout and color contrast are not analyzed. Review an MCP configuration before installing it. Static checks flag settings such as unpinned packages, download-and-execute commands, broad filesystem access and automatic approvals. No configured command runs. Replace credentials with placeholders before submitting; detection and redaction are best effort. Look up an agent's registration, claimed status, verified domain, activity and disputes, where available. Verify an agent or a full OpenGlass record bundle. Record verification checks signatures and consistency, not whether a real-world action happened or who controlled a key. Inspection findings describe matched patterns, not malicious intent. No findings is not a safety guarantee.

How to connect

  1. Claude: Open the OpenGlass listing in the Claude connectors directory and choose Connect. Claude listing OpenGlass exposes 4 tools here.
  2. ChatGPT: Open the OpenGlass listing in the ChatGPT app directory and choose Connect. ChatGPT listing OpenGlass exposes 4 tools here.

Tools

check_mcp_config

ChatGPT
Reads an MCP client config statically and lists settings worth a look before installing it: install-and-run commands (npx, uvx and similar) without an exact version, download-and-execute chains, plain http:// endpoints (loopback and other hosts reported separately), broad filesystem scope, sudo, container or shell entrypoints, auto-approved tools, and credentials in env, headers or args. Accepts {"mcpServers": {...}}, a map of server names to server blocks, or one server block. Credential detection is best-effort and redacts what it finds; replace real values with placeholders before sending. Nothing is looked up, downloaded or run, and nothing is stored. It cannot tell whether a server is malicious, and no findings is not a safety guarantee.

inspect_hidden_content

ChatGPT
Lists content in HTML or plain text that a person looking at it wouldn't see: invisible Unicode (zero-width characters, bidi overrides and isolates, tag characters), text hidden by inline styles or the hidden attribute, HTML comments, large data-* attributes, meta tags, accessibility text that isn't otherwise visible, long base64-looking runs, and an inventory of outbound hosts. Plain text gets the Unicode, encoded-run and reference checks only. One labeled heuristic, capped at 10 matches, flags text that reads like it addresses an AI reader; it is not a judgment of intent or danger. Give exactly one of html, text or url. A url is fetched once with GET: public http/https hosts on standard ports only, at most 2 MB, no subresources, scripts or browser. Hidden-text checks read raw markup and inline styles only, so actual visibility is unknown. Deterministic (same input, same findings); nothing is stored.

lookup_agent

ChatGPT
Realignment R2 (SPEC §14): look up a counterparty BEFORE accepting an invite or offering a session with them — no signing, no auth, free. Pass exactly one of agentId, domain, agentCardUrl, or publicKey. For a registered agent, returns whether it's claimed, its verified domain (if any), activity facts (sessions/attestations in the last 90 days — only counting counterparties that are themselves domain-verified, so these numbers are conservative by design, not a popularity score), open disputes, and flags (new agent, unverified domain, recently rotated key). For an unregistered one, returns whatever public signals exist (its A2A agent card if fetchable, an MCP Registry entry, domain age) plus an invite URL you can pass along. This is informational — decide what an unverified or new counterparty means for your own situation; this tool never tells you to block or refuse anyone.

verify_agent

ChatGPT
Two independent checks, pick one: pass agentId to look up a counterparty's public profile before trusting them (confirms they're registered and claimed — do this before accepting an invite from someone you don't already know). Or pass bundle (a full record bundle, e.g. from get_record with bundle=true) to independently re-run the full cryptographic verification (SPEC §7.6) against OpenGlass's own trusted platform keys — this is a public check, no signing required, and anyone can run the same check themselves, so the record is independently verifiable without trusting OpenGlass's word for it.

check_mcp_config

Claude

inspect_hidden_content

Claude

lookup_agent

Claude
Realignment R2 (SPEC §14): look up a counterparty BEFORE accepting an invite or offering a session with them — no signing, no auth, free. Pass exactly one of agentId, domain, agentCardUrl, or publicKey. For a registered agent, returns whether it's claimed, its verified domain (if any), activity facts (sessions/attestations in the last 90 days — only counting counterparties that are themselves domain-verified, so these numbers are conservative by design, not a popularity score), open disputes, and flags (new agent, unverified domain, recently rotated key). For an unregistered one, returns whatever public signals exist (its A2A agent card if fetchable, an MCP Registry entry, domain age) plus an invite URL you can pass along. This is informational — decide what an unverified or new counterparty means for your own situation; this tool never tells you to block or refuse anyone.

verify_agent

Claude
Two independent checks, pick one: pass agentId to look up a counterparty's public profile before trusting them (confirms they're registered and claimed — do this before accepting an invite from someone you don't already know). Or pass bundle (a full record bundle, e.g. from get_record with bundle=true) to independently re-run the full cryptographic verification (SPEC §7.6) against OpenGlass's own trusted platform keys — this is a public check, no signing required, and anyone can run the same check themselves, so the record is independently verifiable without trusting OpenGlass's word for it.

Similar Apps

Listed on MCP App Store

Listed on MCP App Store

Add this badge to your site or README. It links to this page.

HTML

<a href="https://mcp-app-store.keenethics-labs.com/app/openglass"><img src="https://mcp-app-store.keenethics-labs.com/badges/listed-on-mcp-app-store.svg" alt="Listed on MCP App Store" width="220" height="44"></a>

Markdown

[![Listed on MCP App Store](https://mcp-app-store.keenethics-labs.com/badges/listed-on-mcp-app-store.svg)](https://mcp-app-store.keenethics-labs.com/app/openglass)