This quick guide assumes you have already set up Auto-GPT. If you haven’t, follow our in-depth guide on the Finxter blog.
Use ./run.sh --help
(Linux/macOS) or .\run.bat --help
(Windows) to list command line arguments. For Docker, substitute docker-compose run --rm auto-gpt
in examples.
Common Auto-GPT arguments include: --ai-settings <filename>
, --prompt-settings <filename>
, and --use-memory <memory-backend>
. Short forms such as -m
for --use-memory
exist. Substitute any angled brackets (<>
) with your desired values.
Enable Text-to-Speech using ./run.sh --speak
.
π Use continuous mode (potentially hazardous, may run indefinitely) with ./run.sh --continuous
. Exit with Ctrl+C
.
π§ Use Self-Feedback mode (increases token usage, costs) by entering S
in the input field.
Run GPT-3.5 only mode with ./run.sh --gpt3only
or set SMART_LLM_MODEL
in .env
to gpt-3.5-turbo
. For GPT-4 only, use ./run.sh --gpt4only
(raises API costs).
Find logs in ./output/logs
, debug with ./run.sh --debug
.
Disable commands by setting DISABLED_COMMAND_CATEGORIES
in .env
. For instance, to disable coding features, use:
DISABLED_COMMAND_CATEGORIES=autogpt.commands.analyze_code,autogpt.commands.execute_code,autogpt.commands.gi
.
Okay, this was dry, here’s a more fun article: π
π‘ Recommended: 30 Creative AutoGPT Use Cases to Make Money Online