01
Research integrity
Claims should separate measured results, engineering judgment, and open research direction.
Policies
We maintain shared principles for how persistent agent systems research, infrastructure, and public projects are developed, reviewed, and presented.
These policies define the operating norms for technical rigor, collaboration, attribution, and responsible real-world deployment across the lab.
02
Systems rigor
Benchmarks, interfaces, and operational tradeoffs should remain visible as systems mature.
03
Open collaboration
Work should be reviewable, discussable, and accessible across academic and industry contributors.
04
Attribution
Authorship, code ownership, and licensing should remain explicit across papers, repos, and demos.
05
Responsible deployment
Cost, latency, privacy, safety, and failure modes belong inside the system design conversation.
06
Project stewardship
Lab projects need clear maintainers, visible goals, and a path toward durable public operation.
Implementation Notes
What these policies mean in practice.
- Major technical claims should point to code, benchmarks, experiments, or written records.
- Projects should state ownership, interfaces, and what readiness means before wider deployment.
- Research outputs should preserve attribution across papers, repos, notes, and downstream reuse.
- Operational decisions should include failure handling, cost, latency, and human override paths.