Skip to content
Kanonik
Log inTalk to usStart your trial
Menu
Help
Browse Help
On this page

Connecting your AI

One step and one sign-in connects the AI you already use to your workspace.

Kanonik does not give you an AI to chat with. You bring the AI client you already use and connect it to your workspace. Kanonik gives that AI a set of tools for compliance work, and your AI does the reasoning. Each item it drafts goes through a safety check, then waits for your approval before it reaches your record. Your decisions go into a permanent audit log.

Connecting takes one step and one sign-in. After that your AI stays connected and knows how to do the work.

Open the "Connect your AI" page in the dashboard. It shows the exact connection link for your workspace. Every path below uses that same link, so copy it first.

Path 1: Claude Code

If you work in Claude Code, run this command and replace the last part with your workspace link from the "Connect your AI" page.

claude mcp add --transport http --scope user kanonik <your-workspace-MCP-URL>

The first time you use it, a sign-in opens in your browser. After you sign in, the connection stays in place and you will not be asked again.

Path 2: Claude Desktop

If you work in Claude Desktop, add Kanonik as a custom connector, as the "Connect your AI" page shows:

  1. In Claude, open Customize, then Connectors.
  2. Click + Add, then Add custom connector.
  3. Name it Kanonik, paste your workspace address from the "Connect your AI" page and click Continue.
  4. Keep the sign-in settings Claude suggests and click Add.
  5. Click Connect and sign in to Kanonik in the window that opens. It always asks, so you can see which account connects.

On Claude Team or Enterprise, an Owner adds Kanonik once under Organization settings, then Connectors.

Path 3: Any other MCP-capable client

If your AI client speaks MCP, point it at the same workspace link over streamable HTTP. The connection and sign-in work the same way as the paths above.

What happens after you connect

Kanonik's skills load over the connection on their own. There is nothing to install. These skills are the instructions that turn your AI into a compliance worker, so it already knows how to draft a policy, assess a control, build out your framework, and prepare for an audit. You do not have to teach it the method.

From then on the work happens in your AI client. When your AI proposes a change, the safety check runs, and the proposal waits for you under "To decide" in the dashboard. You decide on each item there, and your decisions are recorded.

Your first prompt

You do not need a special command to begin. Tell your AI what you want in plain language. Two good openers:

  • "Start my ISO 27001:2022 program."
  • "Help me find my risks."

Your AI will use the Kanonik tools, draft the first artifact, and show it to you with the safety-check result. You refine it together, and nothing is recorded until you approve it.

If your AI says Kanonik is unavailable

Sign-ins last about an hour. A client that supports refresh renews its sign-in on its own and you do not notice. Some clients do not, or sometimes lose track of it, and then your AI may report that Kanonik is "unavailable" or "temporarily down" while the service is working normally.

When Kanonik gets a request with no valid sign-in, it answers with a standard "sign in again" response that points the client to the sign-in. A client that follows the standard will restart the sign-in by itself. If yours does not, the fix is a short reconnect:

  1. Remove the Kanonik connection in your AI client's connector or MCP settings and add it again with your workspace link from the "Connect your AI" page, or use the client's own "reconnect" or "re-authenticate" action if it has one.
  2. Sign in when the browser opens. Your records, approvals and skills are unchanged; the connection holds no data of its own.
  3. Ask your AI to read your current state, for example "where do I stand?", to confirm it is working.

If a reconnect does not help, email [email protected] and a person who works on the product answers.

The safety fact that matters

Kanonik never holds the credentials to your own systems. When your AI needs to read your code, your cloud, or your ticketing tool, it does that through its own tools and its own access, not through Kanonik. Kanonik only ever receives what your AI proposes to write into your compliance record, and even that waits for your approval before it lands. Your systems stay yours.

More help

Browse every article in the Help center. For anything else, email [email protected] and a person who works on the product answers.

Kanonik LLC