Common Errors

Quick reference for common errors and their solutions.

---

Authentication Errors

“Invalid token” on sign-in

Symptoms: Signing in with a PAT returns an invalid token error.

Causes:

• Token was mistyped or contains extra whitespace
• Token has been revoked in GitHub
• Token has expired
• Token type is not recognized

Solutions:

1. Copy the token directly from GitHub — avoid retyping
2. Check the token status at GitHub → Settings → Developer settings → Personal access tokens
3. Generate a new token and try again

---

“Sign in with a personal access token” on sign-in

Symptoms: Signing in with a token is rejected with a message asking for a personal access token.

Causes: The token is an OAuth token (gho_, such as the output of gh auth token), a GitHub App user token (ghu_) or an installation token (ghs_, such as a GitHub Actions GITHUB_TOKEN). These are issued to an app rather than created by you, so token sign-in does not accept them.

Solutions:

1. Create a fine-grained or classic personal access token at GitHub → Settings → Developer settings → Personal access tokens and sign in with that
2. Or use Log in with GitHub if your installation has an OAuth app configured

---

“PAT login over non-local HTTP is disabled”

Symptoms: Signing in with a PAT is rejected even though the browser shows HTTPS.

Causes: Your reverse proxy terminates TLS and forwards plain HTTP to the container, so the backend only learns the original scheme from the X-Forwarded-Proto header. Either the proxy is not sending it, or the image predates the fix that stopped the container’s own nginx from overwriting it.

Solutions:

1. Confirm your proxy forwards X-Forwarded-Proto: https — Caddy and Traefik do this automatically; a hand-written nginx proxy needs proxy_set_header X-Forwarded-Proto $scheme;
2. Pull the latest image and restart
3. See HTTPS Setup for per-proxy examples and a stop-gap override

---

“Log in with GitHub” button not appearing

Symptoms: The OAuth login button is missing from the login page.

Causes: GITHUB_CLIENT_ID environment variable is not set or the container was not restarted after adding it.

Solutions:

1. Verify GITHUB_CLIENT_ID is set in your container environment
2. Restart the container to apply the new environment variable
3. Check container logs for OAuth configuration errors

---

“Redirect URI mismatch”

Symptoms: After authorizing the GitHub OAuth App, GitHub returns a redirect URI mismatch error.

Causes: The VITE_BACKEND_URL environment variable does not match the callback URL registered in your GitHub OAuth App.

Solutions:

1. Check the Authorization callback URL in your GitHub OAuth App settings
2. Verify the VITE_BACKEND_URL environment variable matches exactly (including protocol and port)
3. Update the OAuth App or environment variable so they match

---

API and Permission Errors

“Not found” or missing repositories

Symptoms: Repositories are not visible or return 404 errors.

Causes:

• Fine-grained PAT was created with limited repository access
• The GitHub token does not have access to the repository
• The repository is in an organization with restricted token access

Solutions:

1. Check which repositories are authorized on the fine-grained PAT
2. Verify the signed-in GitHub account has access to the repository
3. For organization repositories, check if the organization has blocked token access

---

“Insufficient permissions” on workflow operations

Symptoms: Creating, updating, or deleting workflows fails with a permission error.

Causes: The token is missing required permissions.

Solutions:

1. For fine-grained PATs, verify Contents: Read and write and Actions: Read and write are set
2. For classic PATs, verify the repo and workflow scopes are granted
3. Generate a new token with the correct permissions

---

“More than one project is named …”

Symptoms: An action on a project you don’t own fails with More than one project is named ‘X’. Open it from the projects list, or rename one so they can be told apart.

Causes: Project names are unique per owner, not across the workspace. A project page tells ActionsManager which project it shows, so actions taken from the page act on that project even when another owner’s project has the same name. A request that only names the project — a bookmarked or shared project link, or a direct API call — can’t say which one it means when two owners use that name and you own neither (or hold a grant on both). ActionsManager refuses rather than guess, because a guess could act on the other project.

The owners of each project are not affected: a name always resolves to the project they own.

Solutions:

1. Open the project from the projects list rather than a bookmarked link.
2. Or rename one of the projects from the project sidebar (see Renaming a Project), so links and direct calls work too. Anyone with editor access to that project can rename it, and its owner and workspace admins always can.

---

Workflow and PR Errors

PR campaign creates no PRs

Symptoms: Starting a PR campaign completes without creating any pull requests.

Causes:

• No repositories in the project are configured
• The workflow is already up to date in all repositories
• The GitHub token lacks Pull requests: Read and write permission

Solutions:

1. Verify the project has repositories added
2. Check drift status — if all repositories are in sync, no PRs are needed
3. Verify the PAT has pull request permissions

---

“Branch already exists” when creating campaign PRs

Symptoms: PR creation fails because the branch already exists.

Causes: A previous campaign created a branch that was never merged or deleted.

Solutions:

1. Check the repository on GitHub for open or abandoned campaign branches
2. Delete the old branch and retry the campaign
3. Or merge/close the existing PR first

---

Next Steps

• GitHub Permissions — GitHub permission requirements in detail
• Container Startup — container startup and health issues