FAROS
by OpenNSWM-Lab
Blueprint-driven runtime for orchestrating AI research with agent pipelines
How It Works
Orchestrates end-to-end AI research workflows from ideation and experiments to paper drafting and peer review. Uses blueprint-driven agents and pipelines to compose specialist researcher agents, manage experiments, and stitch outputs into reproducible artifacts. Includes task routing, experiment tracking, and extensible connectors for LLMs and model evaluation tooling. See how the Sub-Agent Delegation Pattern can scale coordination, and how the Model Context Protocol (MCP) Pattern supports consistent context sharing.
The Value Proposition
When to Use
Research teams and organizations automating iterative experiments, paper drafting, and reproducible multi-agent research workflows. The toolkit is well-suited for teams adopting the Blackboard Pattern for collaborative orchestration.
How It's Used
- Automating literature reviews, hypothesis generation, and structured experiment plans
- Running and tracking repeated model experiments with provenance for reproducibility
- Composing specialist researcher agents to draft, edit, and format papers with peer-review loops