autogpt-agents

Clone the AutoGPT repository, start the backend with docker compose, and run the frontend with npm. Then open the visual builder at localhost:3000, add blocks from the BlocksControl panel, connect nodes, configure inputs, and run the agent.

AutoGPT suits continuous agents with visual workflows, triggers, and a marketplace of reusable blocks. LangChain offers more control over agent logic in code, while CrewAI focuses on role-based multi-agent collaboration.

Create a scheduled execution by posting a cron expression with the graph_id and inputs to the platform API, for example "0 */2 * * *" to run every two hours. Agents can also be triggered manually via the execute endpoint or through webhooks.

Stuck executions usually mean the executor service is down or the RabbitMQ queue is backlogged. Check executor logs with docker compose logs executor, inspect queues at the RabbitMQ management UI on port 15672, and restart or scale the executor service.

Yes, add credentials under Profile > Integrations using API keys or OAuth for providers like OpenAI, Anthropic, GitHub, Google, Discord, and Notion. Blocks automatically access stored credentials, which are encrypted at rest.

Avoid AutoGPT when you need fine-grained programmatic control over agent logic, simple hosted assistants, or Microsoft ecosystem integration. In those cases LangChain, OpenAI Assistants, or Semantic Kernel are better fits.