Connect an AI assistant

Once connected, an AI assistant can read Deerlight documents and knowledge and, when permitted, write back to the same project. You no longer need to paste the same background into every conversation.

The quickest way: let it connect itself

Send this sentence to your assistant:

Read https://deerlight.lumenlan.com/skill.md and follow it to connect Deerlight

Three things happen next:

1. It asks for access and gives you a link and an 8-character pairing code, such as K7QM-4TXR.
2. You open the link (signed in to Deerlight), check that the code matches exactly, choose which projects it may see and whether it may write, and choose Allow.
3. The credential goes straight to the assistant. It installs it and reads your projects to confirm.

You never handle a key. The pairing code is valid for 10 minutes; if it expires, ask the assistant to start again.

Only approve a request you just started yourself. If you were not expecting it, or the codes do not match, choose Deny.

Irreversible permissions such as deleting a whole project cannot be requested this way; grant them only when you create a credential by hand.

Create a credential by hand

Some clients cannot request access on their own. Open Connect and create a credential:

Name — say where it is installed, such as “Workstation Claude Code.” You may eventually have several credentials, and a clear name is how you tell them apart.

Access — leave it empty for all projects, or pick one project to limit it to that project and its subprojects.

Permissions — read-only by default. Tick Write documents if the assistant should make changes.

After the credential is created, Deerlight shows the appropriate setup:

• Claude Code and Codex: copy and run one command.
• Other clients: use the general configuration shown on the page.

The credential appears only once and has a dedicated copy action. Treat it as the key to your knowledge base. Put it in client configuration, not in a chat, transcript, or screenshot.

Use one credential per machine

Give each machine its own credential:

• if a machine is retired, you can revoke only that credential;
• the Connect page shows activity per credential, making problems easier to trace.

What an assistant can do

Depending on its permissions, a connected assistant can:

• inspect the project hierarchy;
• read documents and their version history;
• create and update documents;
• create subprojects and shelves;
• obtain skills in its own format — see Skills;
• search documents and approved knowledge.

Revoke and clean up

When a machine no longer needs access, remove its credential under Connect. It stops working immediately. Credentials your assistant requested itself are listed there too, marked as requested by an agent.

The activity history remains available for audit. Delete that history later only when you are sure it is no longer needed.

If you use Kite

You can connect Kite directly from the Connect page. Once linked, sessions opened through Kite can receive Deerlight context without installing a separate credential in every session.