Skip to main content

​
Installation Issues

If the install script fails to download:
  1. Check your internet connection
  2. Verify curl is installed: which curl
  3. Try with verbose output: curl -fsSL -v https://cli.devin.ai/install.sh | bash
If you’re behind a corporate proxy, you may need to configure proxy settings:
export https_proxy=http://your-proxy:port
curl -fsSL https://cli.devin.ai/install.sh | bash
The installer needs write access to install the binary. If you see permission errors:
  1. Check the install location has write permissions
  2. Do not run the installer with sudo — this can cause ownership issues
  3. If installing to a system directory, ensure your user has appropriate permissions
If the install completes but devin isn’t found:
  1. Restart your terminal or run source ~/.bashrc (or ~/.zshrc)
  2. Check if the binary location is in your PATH: echo $PATH
  3. Verify the binary exists: ls -la ~/.local/bin/devin (or the install location shown during setup)

​
Authentication Issues

If browser-based login doesn’t work:
  1. Try the manual token flow for remote/SSH sessions: 
    devin auth login --force-manual-token-flow
  2. Check that your browser can reach the authentication URL
  3. Verify your enterprise account has Devin for Terminal access enabled
If you see authorization errors after logging in:
  1. Verify your account has the the correct permission needed to access Devin for Terminal. You may need to ask your admin. For enterprises, see Devin Auth or Windsurf Auth on how to configure access.
  2. Try logging out and back in: devin auth logout && devin auth login
  3. Check your authentication status: devin auth status
Authentication tokens expire periodically. Simply run:
devin auth login
to refresh your credentials.

​
Runtime Issues

If you see errors about a model not being available:
  1. Check if your enterprise restricts available models in Team Settings
  2. Verify the model name is correct — use /model to see available options
  3. Try a different model: devin --model sonnet -- your prompt
If you hit usage limits:
  1. Wait a few minutes before retrying
  2. Check your organization’s usage dashboard for quota status
  3. Contact your admin if you need higher limits
If the agent stops responding:
  1. Press Ctrl+C to interrupt the current operation
  2. Try /clear to start a fresh session
  3. Check your network connection
  4. Restart Devin for Terminal

​
MCP Server Issues

If an MCP server fails to start:
  1. Verify the command works outside Devin for Terminal: 
    npx -y @modelcontextprotocol/server-github
  2. Check that all required environment variables are set
  3. Look for error messages in the server output
If MCP tools don’t show up:
  1. The server may need a moment to initialize — wait a few seconds
  2. Check that the server is configured correctly in your config file
  3. Verify your enterprise allows MCP servers in Team Settings
MCP tools default to prompting for approval. To auto-approve specific tools, add them to your permissions config:
{
  "permissions": {
    "allow": ["mcp__github__list_issues"]
  }
}

​
Getting Help

If you’re still experiencing issues:
  • Email support: support@cognition.ai
  • Submit a bug report: Use the /bug command inside Devin for Terminal to report issues directly
  • Check for updates: Run devin update to ensure you’re on the latest version