System Requirements
To ensure your AI agents perform efficiently, your host machine (where you run the Burro agent) should meet the following requirements:
- Operating System: macOS (Intel/Apple Silicon), Linux (Ubuntu 20.04+ recommended), or Windows 10/11 (via WSL2).
- CPU: 4+ cores recommended for multi-agent parallel execution.
- RAM: 8GB minimum (16GB+ recommended if running complex development environments).
- Network: Outbound HTTPS (port 443) access to
burros.aiand your chosen LLM provider.
Frequently Asked Questions
#### Getting started
How many Burros do I need?
It depends on the Playbook. A simple task might need just one generalist Burro. A full SDLC mission typically needs 4 Burros covering the Coordinator, Developer, Architect, and QA roles to enable parallel work.
What LLMs are supported?
Burros.AI is LLM-agnostic. It supports OpenAI (GPT-4o), Anthropic (Claude), Google (Gemini), and any OpenAI-compatible local model endpoint.
Can I use my own SSH keys or Git credentials?
Yes. You can add your keys to the Secret Vault or configure them within the Burro's host environment.
#### During missions
Can I run multiple missions at once?
Each project supports one active mission at a time. You can run missions in different projects simultaneously.
What happens if a Burro goes offline during a mission?
Tasks assigned to an offline Burro will wait until it comes back online. Other tasks that don't depend on it continue running normally.
Can agents access the internet?
Yes, agents can browse the web and make API calls — subject to trust policies. Sensitive external operations may require your approval via a Decision Gate.
What happens if an agent fails a task?
The system has a built-in retry mechanism. If it fails repeatedly, the Coordinator will notify you in Mission Control and ask for guidance or a different approach.
#### Data & security
Where does my data live?
Context documents and project artifacts are saved in the portal's secure cloud store. However, all task processing, file manipulation, and code execution happen locally on your infrastructure (where your Burro is running).
Are my secrets safe?
Yes. Secrets are encrypted at rest, injected per-task at runtime, and never logged or displayed in agent telemetry.
Can different teams see each other's projects?
No. Projects are isolated by Corral and organization. Different Corrals enforce strict data boundaries.
Is it safe to give agents shell access?
Agents only run the commands you allow via Decision Gates. You have full visibility and control over every impactful shell command.
#### After missions
Can I reuse deliverables from a previous mission?
Yes. All deliverables are stored in the project Vault and persist across missions. New missions in the same project can reference and build upon previous work via the Blackboard.
How do I export my results?
Deliverables in the Vault can be downloaded individually. For code projects, agents can commit directly to your git repository.
Can I restore an archived project?
Yes. Archived projects can be restored to active status from the project settings.