Harbor Framework (creators of Terminal-Bench)

Harbor

Framework for running and grading agents on benchmark or custom tasks in parallel sandboxes, from the Terminal-Bench team.

Apache-2.0Pythonv0.24.0

What it supports

  • Evaluates multi-step agents and tool calls: supportedEvaluates multi-step agents and tool callsDocs
  • LLM as a judge scoring: supportedLLM as a judge scoringDocs
  • Tracing: supportedTracingDocs
  • ?Runs in CI: not yet verifiedRuns in CI

Install

pip install harbor

Links

Official documentationgithub.com/harbor-framework/harborHosted option: Harbor Hub

Other eval frameworks

Inspect AIpromptfooDeepEvalPydantic EvalsLangfuseOpikMLflow GenAI EvaluationArize PhoenixW&B WeaveRagasGoogle ADK EvaluationBraintrustLangSmith EvaluationMicrosoft Foundry EvaluationsGen AI Evaluation Service

Get the weekly agent stack update

New official MCP servers, spec changes and harness releases, checked against the source. One email a week, no fluff.

Reviewed Oct 9, 2026.