IAS Docs

Add a Repository

Register a Git repository in the IAS Console and link it to your local checkout so agents can operate on your codebase.

Add a Repository

Adding a repository to IAS is a two-part process: register it in the Console, then link it to your local machine. Once connected, you can bootstrap IAS into the repo, build context, and start running agents.

Prerequisites

Before you begin, make sure you have:

  • A workspace in the IAS Console. If you have not created one yet, see Create a workspace.
  • The IAS CLI installed -- install via npm i -g @lucentivelabs/ias or use the framework repo directly.
  • A local Git checkout of the repository you want to connect.

Step 1: Register the repo in the Console

  1. Open the Console and navigate to Repositories in the sidebar.
  2. Click Add Repository.
  3. Enter the clone URL for your repo. Both SSH and HTTPS formats are supported:
    • SSH: git@github.com:your-org/your-repo.git
    • HTTPS: https://github.com/your-org/your-repo.git
  4. Click Create.
  5. Copy the Repo ID that appears (e.g., repos/abc123). You will need this if you configure your worker manually.

At this point, the Console knows about your repository. No code is cloned or stored in the cloud -- only metadata like the URL hash and display name.

Step 2: Install the IAS CLI

If you have not already installed the CLI, run:

npm i -g @lucentivelabs/ias

Then authenticate:

ias auth login

This opens your browser so you can sign in via the Console. The resulting token is stored at ~/.ias/auth.json and reused across all your projects.

You can verify your setup at any time:

ias doctor

This checks configuration, authentication status, control plane connectivity, and repo mappings.

On your local machine, run:

ias repo link /path/to/your/repo

This command:

  1. Detects your remote.origin.url from the local Git checkout.
  2. Matches it against the repos registered in the Console.
  3. Writes a mapping into your worker configuration at ~/.ias/worker.json.

If IAS is not yet installed in the repo, the CLI will offer to install it for you. To skip the install prompt:

ias repo link /path/to/your/repo --install false

Under the hood, ias repo link adds an entry to repos.mappings in your worker config:

{
  "repos": {
    "mappings": [
      {
        "repoId": "repos/abc123",
        "localPath": "/Users/you/code/your-repo"
      }
    ]
  }
}

The worker picks up mapping changes automatically within a few seconds. You can link multiple repos by running ias repo link for each one.

Linking multiple repositories

You can connect as many repositories as you need to a single workspace. Run ias repo link for each one:

ias repo link /path/to/repo-one
ias repo link /path/to/repo-two

Each repo gets its own entry in repos.mappings. Workers will only claim jobs for repos that are explicitly mapped -- this is a safety default that ensures you always control which repos an agent can operate on.

What happens next

Once the repo is registered and linked, the typical next steps are:

  1. Bootstrap IAS -- Deploy the IAS scaffold (docs/ias/, agent configs, CLI tooling) into the repository.
  2. Build context -- Populate the Context Lake with project metadata so agents understand your codebase.
  3. Configure and start agents -- Run a local agent to start processing tasks.
  4. Create goals -- Define what you want built and let agents execute.

Security: what data is shared

IAS is designed to keep your source code on your machine. Here is what the Console stores and what it does not:

Stored in the Console (cloud)Stays on your machine (local)
Repo URL hash and display nameSource code and Git history
Job status, timestamps, retry stateLocal file changes and working tree
Commit SHAs and PR URLs (evidence pointers)The actual commits and pull requests
Decision records and goal metadataIAS scaffold files (docs/ias/)
Agent heartbeat and availabilityAgent execution logs

The Console never clones, reads, or stores your source code. All code changes happen locally on your machine through the agent process. The worker only operates on repos you explicitly map in your configuration.

On this page