Guides

design-research-agents is the execution layer of the CMU Design Research Collective design-research ecosystem. It owns executable participants, workflow and tool runtimes, model-client adapters, and traceable run results.

Primary Path

Follow the same path used across the design-research libraries:

1. Installation — create a supported Python environment and choose extras.

2. Quickstart — run the offline base-install example.

3. Concepts and Typical Workflow — learn the contracts and execution flow.

4. Examples — move from the base example to realistic participants and backends.

5. API — confirm the compatibility-guaranteed import surface.

Additional Guides

• Where To Start helps choose among agents, workflows, patterns, and examples.

• Run An Example In VS Code provides an editor-first setup and debugging path.

• Philosophy explains the package’s research-oriented design choices.

• Dependencies and Extras maps optional backends to install profiles.

Compatibility and Ecosystem

The curated top-level API and documented public facades are the compatibility boundary. design_research_agents.study is preferred for new study runners; design_research_agents.integration remains available for compatibility. See API and Module Reference for the exact distinction.

Sibling layers own adjacent responsibilities:

• design-research-problems owns tasks and evaluators.

• design-research-experiments owns study design and orchestration.

• design-research-analysis owns validation and downstream analysis.