SPIKE-10 Conductor Watch Task Brief

Implement live operator monitoring for tnh-conductor status.

Goal

Add a watch mode so operators can monitor one run until it reaches a terminal lifecycle state without manually rerunning the command.

Example target shape:

tnh-conductor status run-123 --watch

Expected Behavior

• Add --watch to tnh-conductor status.
• Add a configurable polling interval flag.
• When --watch is enabled, print status snapshots repeatedly until the run reaches a terminal state.
• Terminal states should include at least completed, failed, and blocked.
• Preserve the current one-shot status behavior when --watch is not supplied.
• Keep output machine-readable JSON, one snapshot per line.

Likely Files

• src/tnh_scholar/cli_tools/tnh_conductor/tnh_conductor.py
• tests/cli_tools/test_tnh_conductor.py
• optional CLI docs if the implementation naturally reaches them

Validation

At minimum, run:

poetry run pytest tests/cli_tools/test_tnh_conductor.py

Success Criteria

• one-shot status still works unchanged
• status --watch exits once the run reaches a terminal state
• repeated snapshots are emitted while the run is still active
• invalid or missing runs still fail clearly
• tests cover both one-shot and watch behavior