# Harbor

> Framework from the Terminal-Bench creators for evaluating and optimizing agents and models in sandboxed environments, run in parallel across cloud sandbox providers.

- id: `harbor`
- category: Evals & Observability (`evals-observability`)
- url: https://www.harborframework.com/
- repo: https://github.com/harbor-framework/harbor
- status: active
- pricing: open-source
- license: Apache-2.0
- tags: evals, benchmarks, sandbox, agents
- related: Terminal-Bench (https://indexagentica.com/entries/terminal-bench/); Daytona (https://indexagentica.com/entries/daytona/); Modal (https://indexagentica.com/entries/modal/); Blaxel (https://indexagentica.com/entries/blaxel/)
- sources: https://github.com/harbor-framework/harbor
- added: 2026-10-21
- updated: 2026-10-21
- last verified: 2026-10-21
- maintainer: Harbor (Terminal-Bench team)
- submitted by: agentica-curator
- page: https://indexagentica.com/entries/harbor/
- json: https://indexagentica.com/api/entries/harbor.json

Evaluates agents such as Claude Code, OpenHands and Codex CLI; build and share benchmarks and environments; providers include Daytona, Modal and Blaxel.