Post the agent's duration with the run

Post the agent's duration with the run

Sums each trial's agent_execution span and sends duration_seconds, so
the Gym's new Duration column fills for every future run — duration is
the local lane's whole cost.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01GoYpb8FEmdxVErsv7ABCYi

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01GoYpb8FEmdxVErsv7ABCYi

Deploy story

What this commit did to the running system — joined from the forge receipt chain, the part a commit page elsewhere cannot show.

Not deployed through the forge lane

No push, promotion, build, or deploy receipt references this commit (receipts are scanned over a bounded recent window). Changes shipped by full node replacement carry their proof in the release gate receipt instead.

Changed files

modified bench/post_gym_run.py

Diff

1 file changed, +10 -0

    trials = []
    input_tokens = 0
    output_tokens = 0
    duration_seconds = 0.0
    agent_version = None
    model = None
    for trial_dir in sorted(job_dir.iterdir()):

            continue
        trial = json.loads(trial_result.read_text())
        trials.append(trial)
        execution = trial.get("agent_execution") or {}
        started = execution.get("started_at")
        finished = execution.get("finished_at")
        if started and finished:
            from datetime import datetime

            span = datetime.fromisoformat(finished.replace("Z", "+00:00")) - datetime.fromisoformat(started.replace("Z", "+00:00"))
            duration_seconds += span.total_seconds()
        trajectory_path = trial_dir / "agent" / "trajectory.json"
        if trajectory_path.exists():
            trajectory = json.loads(trajectory_path.read_text())

        "tasks_passed": tasks_passed,
        "input_tokens": input_tokens or None,
        "output_tokens": output_tokens or None,
        "duration_seconds": int(duration_seconds) or None,
        "recipe_digest": recipe_digest,
        "report": {
            "job_id": result.get("id"),

48	48		`trials = []`
49	49		`input_tokens = 0`
50	50		`output_tokens = 0`
	51	+	`duration_seconds = 0.0`
51	52		`agent_version = None`
52	53		`model = None`
53	54		`for trial_dir in sorted(job_dir.iterdir()):`

56	57		`continue`
57	58		`trial = json.loads(trial_result.read_text())`
58	59		`trials.append(trial)`
	60	+	`execution = trial.get("agent_execution") or {}`
	61	+	`started = execution.get("started_at")`
	62	+	`finished = execution.get("finished_at")`
	63	+	`if started and finished:`
	64	+	`from datetime import datetime`
	65	+
	66	+	`span = datetime.fromisoformat(finished.replace("Z", "+00:00")) - datetime.fromisoformat(started.replace("Z", "+00:00"))`
	67	+	`duration_seconds += span.total_seconds()`
59	68		`trajectory_path = trial_dir / "agent" / "trajectory.json"`
60	69		`if trajectory_path.exists():`
61	70		`trajectory = json.loads(trajectory_path.read_text())`

98	107		`"tasks_passed": tasks_passed,`
99	108		`"input_tokens": input_tokens or None,`
100	109		`"output_tokens": output_tokens or None,`
	110	+	`"duration_seconds": int(duration_seconds) or None,`
101	111		`"recipe_digest": recipe_digest,`
102	112		`"report": {`
103	113		`"job_id": result.get("id"),`