GOVENANT
Free · hosted · read-only

Run the audit on your own agents.

Connect the GOVENANT Audit to the AI tool you already use. Your agent runs a read-only, aggregate-only sweep of your own system and hands you a plain-English report — is it actually delivering, and can you prove it? Your prompts, data, and secrets never leave your machine.

Add it to your tool → Questions? Read the FAQ

Works with your AI tool — click yours for a step-by-step guide

Claude ChatGPT Claude Code Cursor VS Code Windsurf Codex Cline

Already ran it before? Sign in to your dashboard →

Step 1

Add the audit

Paste one URL into your AI tool. It connects itself — no API key.

Step 2

Sign in (this is your signup)

A browser opens; continue with GitHub, Google, Microsoft, LinkedIn, or GitLab. No separate signup — that sign-in creates your account & dashboard.

Step 3

Ask for the audit

Type: “Run a GOVENANT trust sweep.” You get a shareable report back.

One dashboard for every project

Your sign-in isn't just for one run. Come back anytime to see every project you've audited and its journey over time — how each system climbs Logged → Gated → Delivered → Earned.

Sign in to your dashboard →

Add it to your tool

The server URL is the same everywhere:

https://mcp.govenantstandard.org/mcp

Prefer a dedicated step-by-step guide? Claude · ChatGPT · Claude Code · Cursor · VS Code · Windsurf · Codex · Cline

Easiest — no files, just the app

Claude (desktop app)
  1. Open Settings → Customize → Connectors.
  2. Click + Add custom connector, paste the URL above, click Add.
  3. Click Connect — a browser opens; sign in with GitHub or Google.
  4. In a chat, say “Run a GOVENANT trust sweep.”

If it doesn’t appear, fully quit Claude (⌘Q) and reopen.

ChatGPT
  1. Settings → Apps & Connectors → Advanced → turn on Developer mode.
  2. Apps & Connectors → + Create. Name it “GOVENANT”, paste the URL, set Authentication: OAuth.
  3. Create → authorize in the browser.
  4. Ask it to “Run a GOVENANT trust sweep.”

Developer mode is available on Plus, Pro, Business, and Enterprise.

In your coding tool

Claude Code (CLI) 
claude mcp add --transport http govenant https://mcp.govenantstandard.org/mcp

Then run /mcp inside Claude Code to sign in. Add --scope user to enable it in every project.

Cursor

Add to ~/.cursor/mcp.json, then toggle the server on in Settings → MCP & Integrations:

{ "mcpServers": { "govenant": { "url": "https://mcp.govenantstandard.org/mcp" } } }

Cursor opens the sign-in automatically on first connect.

VS Code (Copilot agent mode)

Add to .vscode/mcp.json (note the key is servers, not mcpServers):

{ "servers": { "govenant": { "type": "http", "url": "https://mcp.govenantstandard.org/mcp" } } }

Requires VS Code 1.102+. Sign-in opens in the browser.

Windsurf

Add to ~/.codeium/windsurf/mcp_config.json (remote servers use serverUrl), then hit Refresh:

{ "mcpServers": { "govenant": { "serverUrl": "https://mcp.govenantstandard.org/mcp" } } }
Codex CLI 
codex mcp add govenant --url https://mcp.govenantstandard.org/mcp
codex mcp login govenant

On older builds, npm i -g @openai/codex@latest first.

Cline

Cline’s sign-in for remote servers is finicky, so use the bridge in cline_mcp_settings.json:

{ "mcpServers": { "govenant": { "command": "npx", "args": ["-y", "mcp-remote", "https://mcp.govenantstandard.org/mcp"] } } }

What the report tells you

Anomalies first

The bad news up top: motion without delivery, missed work, approvals expiring unapproved, decisions never graded — the signs of performed autonomy.

Your grade

Where you land on the GOVENANT ladder (Logged → Gated → Delivered → Earned), and the delivery numbers behind it.

Delivery, not activity

Verified outcomes vs. total actions, work delivered vs. missed — measured against the record, never self-reported.

A shareable page

Every run has its own URL, with an optional one-click “share for a free expert review.”

Questions, in plain English

What is this, really?
A health check for your AI agents. It looks at whether your agents are actually getting work done — or just looking busy — and gives you a simple report card. Think of it like a credit report, but for how well your AI system is governed.
Do I need to be technical?
To connect it once, whoever set up your AI agents does a two-minute setup (paste a URL, sign in). After that, anyone can just ask the assistant “run the audit” and read the report — the report is written in plain English.
What does it look at — and what does it NOT see?
It reads summary numbers from your agents’ own records: how many ran, how many finished, whether checks passed, which AI models you use. It never reads your prompts, your agents’ messages, your customers’ information, your database passwords, or any secrets.
How long does it take?
About two minutes to connect the first time. The audit itself runs in seconds.
Is it really free? What’s the catch?
Yes — free. You sign in with GitHub or Google so we know who you are, and that’s the whole deal. If your report turns up problems, you can optionally ask for a free human review from the team behind the standard. No credit card, no paywall on the audit.
Where does my data go?
Only the summary numbers leave your machine, and only to build your report. The detailed records stay with you. We use anonymized, pooled results to improve the open standard — never anything that identifies you. Full detail here.
My setup is custom — will it work?
Yes. Your own AI assistant figures out how to read your system; you don’t configure anything. It maps your setup to the audit automatically.
What do I get at the end?
A shareable report page: your grade, the top problems it found (worst first), and specific things to fix — plus the option to have an expert review it with you.

What never leaves your machine

The audit is read-only and aggregate-only: counts, ratios, timestamps, model names. We never see your prompts, your agents’ messages, your customers’ data, your credentials, or your secrets — only how your agents are built and whether they work.

Add it to your tool → See it audited on itself