Skip to content
For Self-Managed AI Platform users running v11.1, see the on-premise platform documentation
Developer documentation > DataRobot Agent Assist > Troubleshooting

Troubleshooting

This page describes how to resolve common issues with DataRobot Agent Assist, including plugin discovery, dependency checks, authentication and API keys, template bootstrap failures, LLM API errors, and session interruption. Follow the sections below for step-by-step fixes.

Plugin not discovered

If dr assist does not run or the plugin does not appear in dr plugin list:

Check that dr-assist is on the PATH

which dr-assist

If it is not found, reinstall the plugin and ensure the directory containing dr-assist is on the PATH.

Check manifest output

The CLI discovers plugins by running the executable with --dr-plugin-manifest and reading JSON from stdout.

dr-assist --dr-plugin-manifest

This should print valid JSON. If it hangs or errors, the plugin may not start correctly.

Check manifest speed

The manifest must respond within 100 ms for reliable discovery; the CLI uses a 500 ms per-manifest timeout.

time dr-assist --dr-plugin-manifest

If it is slow, look for heavy imports or startup work before the manifest is printed.

Debug with the CLI

dr --debug plugin list

Use this to see why a plugin might not be listed.

Common causes:

The manifest responds in over 500 ms, the executable lacks execute permission (chmod +x), or the name conflicts with a built-in command.

Dependency check failed

At startup, DataRobot Agent Assist runs dr dependencies check, which uses .datarobot/cli/versions.yaml. If the file does not exist, the application creates it with default minimum versions (Python 3.10, Node 24, git 2.30, task 3.43.3, pulumi 3.163.0).

  • Error panel: If the check fails, the application shows a "Dependency Check Failed" panel and prints the dependency error output.

  • What to do: Install or upgrade the missing tools to at least the versions in the table in Prerequisites and installation. Ensure each tool is on the PATH and reports at least the minimum version (for example, git --version, task --version). Fix any issues reported in the panel (for example, wrong executable name or path).

Authentication / API key errors

DataRobot Agent Assist requires a valid DataRobot API token. Use the following for missing or invalid credentials.

Missing API key

If DATAROBOT_API_TOKEN (or equivalent from configuration) is not set, the application shows a "Configuration Error" panel before starting the chat. Do one of the following:

  • Set the environment variable: export DATAROBOT_API_TOKEN='your-api-key-here'
  • Create or update a .env file in the current directory with DATAROBOT_API_TOKEN=...
  • Get the API key from the URL returned by get_api_key_url() (for example, DataRobot account profile).

Invalid or expired key

On authentication failure (for example, OpenAI AuthenticationError), the application shows an "Authentication Error" panel. Verify the key, run echo $DATAROBOT_API_TOKEN, update .env if needed, and fetch a new key from the provider URL.

Using the DataRobot CLI for authentication

Log in with the DataRobot CLI; the token is stored in the DR CLI configuration and reloaded by DataRobot Agent Assist:

dr auth login

If login fails, the application prints "DataRobot CLI authentication failed" and suggests running dr auth login manually.

Template bootstrap failures

Run dr assist in an empty directory

Only run dr assist from a dedicated and empty directory. Running this command in a directory containing code or other files is unsafe. When you use the agent assist coding workflow, the assistant clones the DataRobot Agent Application Template repository into the current directory. This action can overwrite or conflict with existing files, damaging the existing project and degrading the accuracy of the assistant's output. Before running dr assist, if you're not in a dedicated directory, create one and open the terminal there (for example, mkdir my-agent && cd my-agent, then run dr assist).

If you see unexpected files, overwrites, or odd behavior after a clone, you may have started in a non-empty directory.

If a git clone fails during the coding workflow (for example, when cloning the DataRobot agent template):

Verify git is installed and on the PATH:

git --version

Check network access to the repository:

git ls-remote https://github.com/datarobot-community/datarobot-agent-application.git

If using SSH, verify SSH keys are configured:

ssh -T git@github.com

LLM API errors

For timeouts, rate limiting, or model-not-available errors:

  • Check DataRobot service status.
  • Verify your account has LLM gateway access.
  • Try a different model if the selected one is unavailable.
  • For timeouts, the assistant prompts you to retry.

Session interruption

If the session ends unexpectedly:

  • Session state is not persisted between runs.
  • The agent_spec.md file is saved to disk and preserved.
  • Template directory contents are preserved.
  • Restart the session to continue.

Getting help

For more assistance see the following related documentation:

  • DataRobot CLI: For CLI installation, configuration file format, and authentication flow: DataRobot CLI documentation.
  • Plugin build and CI: See the plugin README in the repository for distribution, build, and troubleshooting.