# kanonik [Health: Active]

**Category:** ⚖️ Legal  
**Repository:** https://github.com/Kanonik-LLC/kanonik-plugin  
**GitHub Stars:** 0  
**Views:** 0  
**Installs:** 0  
**Upvotes:** 0  
**Directory Page:** https://allmcps.com/mcp/kanonik

## Description
Governance runtime for compliance: verified, human-approved writes to a tamper-evident record.

## Claude Desktop Quick Installation
Remote MCP endpoint (confidence: high). Install path detected from listing signals. Add as a URL/SSE server in your client:

```json
"mcpServers": {
  "kanonik": {
    "url": "https://app.kanonik.ai/mcp"
  }
}
```

## Documentation

## What the kanonik MCP server does

The kanonik MCP server gives an AI a structured workspace for compliance activities. It supports drafting and revising policies and procedures against a loaded framework, evaluating internal controls, recording evidence for judgements, identifying gaps, and building an audit trail that an assessor can review. It can also produce a Statement of Applicability or a comparable artifact for the selected framework.

The framework is selected through the workspace rather than fixed in the plugin. The README identifies ISO 27001:2022, SOC 2, GDPR, NIST CSF 2.0, and HIPAA as supported frameworks. The plugin is not described as being specific to any one framework.

## How it works

The AI performs the reasoning through MCP, while Kanonik supplies the compliance model and server-side controls. A Verifier checks the AI's work, and actions are recorded in a tamper-evident record. Proposed changes use the `kanonik:propose` scope, while reading the compliance record uses `kanonik:read`.

A write cannot change the record immediately. It requires a signed, single-use approval token and a human approval action. The AI cannot approve its own work. Committing a change uses the `kanonik:commit` scope.

Kanonik does not manage credentials for external source systems. If the AI accesses GitHub, Jira, or an identity provider, that access comes through the AI client's own connections rather than through Kanonik.

## Setup and configuration

Install the plugin from the Grok marketplace with `/marketplace` and select Kanonik, or run `grok plugin install kanonik --trust` from the CLI. On first use, the client connects to `https://app.kanonik.ai/mcp` and opens a browser for sign-in. A Kanonik account is required.

The plugin contains only an MCP connection manifest. Its configuration points the `kanonik` server name at the remote HTTP endpoint:

```json
{
  "mcpServers": {
    "kanonik": { "type": "http", "url": "https://app.kanonik.ai/mcp" }
  }
}
```

Authentication uses OAuth 2.1 with Dynamic Client Registration and PKCE against `https://auth.kanonik.ai`. Tokens remain with the Grok client. The plugin does not contain credentials or secrets. The documented network endpoint for MCP access is `https://app.kanonik.ai/mcp`.

## Tools and capabilities

The kanonik MCP server supports these compliance tasks:

- Draft and revise policies and procedures against a loaded framework.
- Assess internal controls and attach evidence to each judgement.
- Compare framework requirements with existing materials to identify gaps.
- Maintain a tamper-evident audit record for independent review.
- Produce a Statement of Applicability or an equivalent framework artifact.
- Verify AI-produced work before a proposed change reaches approval.

## Limitations and notes

This repository does not include a local runtime, install script, hook, or bundled implementation. The service runs server-side at Kanonik.

Every write remains subject to human approval, so the server is not suitable when an agent must make unattended record changes. Kanonik also does not directly hold or use credentials for GitHub, Jira, or identity providers. Access to those systems depends on the AI client's separate connections.

_Full upstream README: https://allmcps.com/mcp/kanonik/readme_