This guide walks you through everything you need to go from a new account to a governed AI coding agent. The entire flow takes about five minutes and requires no code changes in your agent or repositories.Documentation Index
Fetch the complete documentation index at: https://docs.runaegis.co/llms.txt
Use this file to discover all available pages before exploring further.
Create your account
Go to app.runaegis.co and sign up. You can register with your email address and a password, or continue with Google or GitHub OAuth, all three options are available on the sign-up screen.After signing up with email, check your inbox for a verification link. You must verify your email before you can log in.
Connect your GitHub account
The first onboarding step asks for three pieces of information:
- GitHub username, your GitHub handle (for example,
octocat). - GitHub user ID, the numeric ID associated with your account. You can find it by visiting
https://api.github.com/users/YOUR_USERNAMEand reading theidfield. - Personal access token (PAT), a classic or fine-grained token with the following scopes:
repo, full repository accessread:user, read your profile informationworkflow, update GitHub Actions workflow files
Sync your repositories
After saving your GitHub credentials, Aegis uses your PAT to discover the repositories you own or have access to. Click Sync repositories to fetch them.Once the sync completes, you will see a list of repositories. All repositories default to Allow at this stage, you can change individual permissions in the next step.
Set repository permissions
For each repository you can choose one of three permission levels:
You can apply a permission level to all repositories at once using the bulk controls, then override individual repos as needed. You can change these settings at any time from Settings inside the dashboard.
| Permission | What it means |
|---|---|
| Allow | The agent can read and write freely. Actions execute immediately without review. |
| Require Approval | The agent can read but all write actions are paused for a human to approve or deny from the dashboard. |
| Deny | All agent access to this repository is blocked. The agent receives an error if it attempts any tool call targeting this repo. |
Connect your agent
Copy the MCP configuration snippet and paste it into your agent’s config file. The snippet uses your GitHub user ID as the Replace
user_id header so Aegis can route requests to the correct account.You must create a room in Aegis before connecting an agent in production. Rooms scope governance to a specific repository and generate the
room_id and access_token values used in the full integration config. See the Rooms documentation for details. During initial onboarding, the snippet below is all you need to verify connectivity.- Claude Code
- Cursor
- VS Code Copilot
- Other
Add the following to Then restart Claude Code.
~/.claude/mcp.json (or .claude/mcp.json at the project level):~/.claude/mcp.json
YOUR_GITHUB_USER_ID with the numeric GitHub user ID you provided in step 2.Verify the connection
After pasting the config and restarting your agent, Aegis waits for the first inbound tool call. The onboarding screen shows a Waiting for first agent action indicator that polls every 5 seconds. As soon as your agent hits any MCP tool, the indicator turns green and confirms the connection is working.If you want to skip verification for now, you can click Skip for now and finish setup. You can always confirm connectivity later from the dashboard.