{% extends "layout.html" %} {% block title %}{{ task_id }} · run {{ run_id }} · cua-speedrun{% endblock %} {% block content %}

{{ task_id }} · seed {{ seed }}

Task-level diagnostics. Back to run {{ run_id }}{% if summary %} · {{ "pass" if summary.passed else "fail" }} {% if summary.task_time_sec is not none %}{{ "%.2f" | format(summary.task_time_sec) }}s{% endif %} {{ summary.num_steps }} steps · {{ summary.num_observes }} observes {% endif %}

{% if instruction %}

Instruction: {{ instruction }}

{% endif %} {% if verdict_detail %}

Verdict: {{ verdict_detail }}

{% endif %} {% if summary %}

Time split

task time (armed → done){{ "%.2f" | format(summary.task_time_sec or 0) }}s
environment time (steps + observes){{ "%.2f" | format(summary.env_time_sec or 0) }}s
agent time (thinking + network){{ "%.2f" | format(summary.agent_time_sec or 0) }}s
{% endif %} {% if phases %}

Preparation phases (untimed)

{% for k, v in phases.items() %}{% endfor %}
{{ k }}{{ v }}
{% endif %} {% if logs %}

Logs

{% endif %}

Trajectory

{% for l in logs %} {% if l.name == "agent.stdout" or l.name == "agent.stderr" %}
{{ "Agent stderr" if l.name == "agent.stderr" else "Agent stdout" }} {{ l.name }} · full captured stream
Open to load.
{% endif %} {% endfor %} {% if not has_frames %}

No screenshots were captured for this task.

{% endif %}
    {% for s in steps %} {% if s.kind == "observe" %}
  1. observe{% if s.dur %} · {{ "%.2f" | format(s.dur) }}s{% endif %}
    {% if s.frame %} screenshot {% else %}no frame{% endif %}
  2. {% elif s.kind == "step" %}
  3. ↳ act{% if s.dur %} · {{ "%.2f" | format(s.dur) }}s{% endif %} {{ s.actions | tojson }}
  4. {% elif s.kind == "done" %}
  5. agent marked done
  6. {% elif s.kind == "verdict" %}
  7. {{ "pass" if s.passed else "fail" }} {{ s.detail }}
  8. {% endif %} {% endfor %}
{% endblock %}