Track experiments from your own optimization loop (LLM agents, custom scripts, manual experiments) in the Weco dashboard.

# Initialize a run (prints run ID to stdout)
WECO_RUN_ID=$(weco observe init --name "my-experiment" --metric val_bpb --goal min --source train.py)

# Log experiment results
weco observe log --run-id "$WECO_RUN_ID" --step 0 --description "baseline" --metrics '{"val_bpb": 2.36}' --source train.py
weco observe log --run-id "$WECO_RUN_ID" --step 1 --description "increase batch size" --metrics '{"val_bpb": 2.26}' --source train.py
weco observe log --run-id "$WECO_RUN_ID" --step 2 --status failed --description "OOM" --metrics '{"val_bpb": 0.0}' --source train.py

Failures are printed to stderr. weco observe init exits non-zero on any failure, so a captured WECO_RUN_ID is always valid. weco observe log exits non-zero only for errors you must fix (bad flags, unreadable files, not logged in, a rejected request); Weco-side blips are retried, then reported without failing the command, so they never crash your agent's loop. Pass --strict to log to make those fatal too. For full documentation, see Weco Observe.

Tracking External (Non-Weco) Runs