AutoGPT Agent User Guide¶
This guide assumes you are in the
autogpts/autogpt folder, where the AutoGPT Agent
Command Line Interface¶
./autogpt.sh (or any of its subcommands) with
--help lists all the possible
sub-commands and arguments you can use:
$ ./autogpt.sh --help Usage: python -m autogpt [OPTIONS] COMMAND [ARGS]... Options: --help Show this message and exit. Commands: run Sets up and runs an agent, based on the task specified by the... serve Starts an Agent Protocol compliant AutoGPT server, which creates...
For Windows users
On Windows, use
.\autogpt.bat instead of
Everything else (subcommands, arguments) should work the same.
Usage with Docker
For use with Docker, replace the script in the examples with
docker compose run --rm auto-gpt:
docker compose run --rm auto-gpt --ai-settings <filename> docker compose run --rm auto-gpt serve
run – CLI mode¶
run sub-command starts AutoGPT with the legacy CLI interface.
./autogpt.sh run --help
$ ./autogpt.sh run --help Usage: python -m autogpt run [OPTIONS] Sets up and runs an agent, based on the task specified by the user, or resumes an existing agent. Options: -c, --continuous Enable Continuous Mode -y, --skip-reprompt Skips the re-prompting messages at the beginning of the script -C, --ai-settings FILE Specifies which ai_settings.yaml file to use, relative to the AutoGPT root directory. Will also automatically skip the re-prompt. -P, --prompt-settings FILE Specifies which prompt_settings.yaml file to use. -l, --continuous-limit INTEGER Defines the number of times to run in continuous mode --speak Enable Speak Mode --debug Enable Debug Mode --gpt3only Enable GPT3.5 Only Mode --gpt4only Enable GPT4 Only Mode -b, --browser-name TEXT Specifies which web-browser to use when using selenium to scrape the web. --allow-downloads Dangerous: Allows AutoGPT to download files natively. --skip-news Specifies whether to suppress the output of latest news on startup. --install-plugin-deps Installs external dependencies for 3rd party plugins. --ai-name TEXT AI name override --ai-role TEXT AI role override --constraint TEXT Add or override AI constraints to include in the prompt; may be used multiple times to pass multiple constraints --resource TEXT Add or override AI resources to include in the prompt; may be used multiple times to pass multiple resources --best-practice TEXT Add or override AI best practices to include in the prompt; may be used multiple times to pass multiple best practices --override-directives If specified, --constraint, --resource and --best-practice will override the AI's directives instead of being appended to them --help Show this message and exit.
This mode allows running a single agent, and saves the agent's state when terminated. This means you can resume agents at a later time. See also agent state.
For legacy reasons, the CLI will default to the
run subcommand when none is
./autogpt.sh run [OPTIONS] does the same as
but this may change in the future.
💀 Continuous Mode ⚠️¶
Run the AI without user authorization, 100% automated. Continuous mode is NOT recommended. It is potentially dangerous and may cause your AI to run forever or carry out actions you would not usually authorize. Use at your own risk.
To exit the program, press Ctrl+C
serve – Agent Protocol mode with UI¶
serve, the application exposes an Agent Protocol compliant API and serves a
frontend, by default on
./autogpt.sh serve --help
$ ./autogpt.sh serve --help Usage: python -m autogpt serve [OPTIONS] Starts an Agent Protocol compliant AutoGPT server, which creates a custom agent for every task. Options: -P, --prompt-settings FILE Specifies which prompt_settings.yaml file to use. --debug Enable Debug Mode --gpt3only Enable GPT3.5 Only Mode --gpt4only Enable GPT4 Only Mode -b, --browser-name TEXT Specifies which web-browser to use when using selenium to scrape the web. --allow-downloads Dangerous: Allows AutoGPT to download files natively. --install-plugin-deps Installs external dependencies for 3rd party plugins. --help Show this message and exit.
For more information about the API of the application, see agentprotocol.ai.
Most arguments are equivalent to configuration options. See
for all available configuration options.
Replace anything in angled brackets (<>) to a value you want to specify
Here are some common arguments you can use when running AutoGPT:
Run AutoGPT with a different AI Settings file
./autogpt.sh --ai-settings <filename>
Run AutoGPT with a different Prompt Settings file
./autogpt.sh --prompt-settings <filename>
There are shorthands for some of these flags, for example
./autogpt.sh --help for more information.
The state of individual agents is stored in the
data/agents folder. You can use this
in various ways:
- Resume your agent at a later time.
- Create "checkpoints" for your agent so you can always go back to specific points in its history.
- Share your agent!
Agents can read and write files. This happens in the
workspace folder, which
data/agents/<agent_id>/. Files outside of this folder can not be accessed by the
RESTRICT_TO_WORKSPACE is set to
We do not recommend disabling
RESTRICT_TO_WORKSPACE, unless AutoGPT is running in
a sandbox environment where it couldn't do any damage (e.g. Docker or a VM).
Activity, Error, and Debug logs are located in
Do you notice weird behavior with your agent? Do you have an interesting use case? Do you have a bug you want to report? Follow the step below to enable your logs. You can include these logs when making an issue report or discussing an issue with us.
To print out debug logs:
Disabling Command Categories¶
If you want to selectively disable some command groups, you can use the
DISABLED_COMMAND_CATEGORIES config in your
.env. You can find the list of available
For example, to disable coding related features, set it to the value below: