Skip to content
Agent Benchmarking · Lab-direct · benchmark engineering

Terminal-agent benchmark built end to end, from task taxonomy to grading pipeline

Agentic evaluation is where crowd models break. Task authoring, Dockerised environment engineering, verified gold solutions and step-level trajectory grading, run as one accountable workstream.

Agent trajectoryGraded step by step, not pass or fail

A run that ends green can still be wrong in the middle. Scoring the trajectory is what separates an agent benchmark from a unit test.

Engagement pattern

A documented program shape with illustrative volumes, not a named client reference.

The problem

The team needed a benchmark that would survive external scrutiny. Reproducible environments, unambiguous success criteria, and grading that two independent engineers would agree on.

How it runs

  1. 1

    Taxonomy first

    Capability dimensions and difficulty bands defined before any task was authored, so coverage was a design decision rather than an artifact of what was easy to write.

  2. 2

    Environment engineering

    Each task ships as a pinned Docker image with a deterministic test harness. A task is not accepted into the suite until a gold solution passes the harness from a clean container.

  3. 3

    Adversarial review

    Every task is attacked by a second engineer looking for shortcut solutions, ambiguous success criteria and environment leakage before it enters the suite.

  4. 4

    Trajectory grading

    Step-level rubric scoring of agent trajectories with agreement tracked between graders, so the benchmark reports capability rather than grader variance.

Outcome

Reproducibility
100%
Gold solution passes from clean container
Grader agreement
κ ≥ 0.7
Tracked per rubric dimension
Shortcut audit
Every task
Adversarially reviewed before inclusion
Handover
Full
Harness runs in your CI, not ours

What makes it stick

Benchmark programs are handed over complete. The harness runs in your infrastructure. We are not a dependency in your evaluation loop.