Apps · PROJECT NOTES
Harbor
An execution framework for agent evaluations and training rollouts, with reusable task formats and local or remote sandbox backends.
What can you do with it?
Run the same agent across selected task datasets, or collect trajectories for subsequent reinforcement learning.
What should you know before using it?
Harbor is the runner used by Terminal-Bench, not a separate task benchmark. Current code requires Python 3.12+. Local runs need Docker; remote providers and model endpoints need their own configuration. Dataset licenses remain separate from Apache-2.0 framework code.
What is verified here?
This entry was reviewed against original project material. We have not independently run the software or reproduced its results. AI-development claims are author-reported; check upstream for current versions, setup requirements, and licenses.
- Resource type
- Environment framework
- Provenance
- Development provenance not established
- Reviewed
- Platforms
- Consult upstream documentation