Running Evaluations

Use ct run eval to run a policy or control protocol against one or more tasks. Start with a test-policy smoke run before spending model tokens:

uv run ct run eval --policy test --task-file simple --tags test

Running a full sabotage evaluation? Use ct run sabotage-eval instead — it orchestrates honest evals, attack evals, monitoring, and safety metrics in a single command.

Guides

GuideWhat it covers
Setup and configurationEnvironments, credentials, models, policies, and external sandboxes
Task selectionExact tasks, task files, trajectories, replay, and expansion modes
ExamplesHonest, attack, batch, and live-development runs
Config filesYAML configuration, composition, and the current schema
EC2 fleet operationWorker AMIs and streaming fleet events
OutputEval logs, Inspect's viewer, and trajectory conversion
RunsUploading, listing, inspecting, sampling, and downloading runs
TroubleshootingDocker address pools and container service startup
ct run eval CLI referenceEvery current argument and option, generated from Click

See Also