Harbor integration
Select Promigence in Harbor with one flag. Terminal-Bench and other Harbor tasks then run in cloud sandboxes, each trial in its own copy of the task's environment and many trials at once, with no change to Harbor.
The package is promigence-harbor on PyPI, for Python.
Install
Install it and set your key.
pip install promigence-harbor
export PROMIGENCE_API_KEY=...
export PROMIGENCE_SPEND_CAP=5
export HARBOR_TELEMETRY=0 # optional: Harbor sends usage telemetry unless you turn it off
Example
From the guide Run Harbor tasks in a Promigence sandbox, which goes on to the spend cap, cleaning up, and what the package does not support.
harbor run -p path/to/dataset -a claude-code -m anthropic/claude-sonnet-4-5 \
-e promigence_harbor:PromigenceEnvironment \
--ek spend_cap=5 --ek sandbox_timeout_sec=3600 -n 32