Hawk

Hawk is a platform for running Inspect AI evaluations at scale on cloud infrastructure. You describe the tasks, agents, and models to evaluate in a YAML config, and Hawk handles the rest: provisioning isolated Kubernetes pods, managing LLM API credentials, streaming logs, importing results into a PostgreSQL warehouse, and serving a web UI to browse them. It also runs Inspect Scout scans over transcripts from completed evaluations.

📚 Full documentation: hawk.metr.org

Install the CLI

Install from PyPI with uv (or pip):

uv tool install "hawk[cli]"   # or: pip install "hawk[cli]"

See the installation guide for pinning to a deployment and installing from source.

Run your first eval

export HAWK_API_URL=https://api.hawk.example.com   # your Hawk deployment
hawk login
hawk eval-set my-eval-set.yaml   # write your own, or grab an example from the repo's hawk/examples/
hawk logs -f    # watch it run
hawk web        # open results in the browser

See Running Evaluations for a ready-to-run example config.

Documentation

Installation — install the CLI and run your first eval
Running Evaluations — eval set configs, secrets, local runs
Running Scans — Inspect Scout scans over transcripts
CLI Reference — full command reference
Deploy your own Hawk — stand up a Hawk instance on AWS
Contributing — developer setup and testing

License

MIT

Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

hawk-2.4.0.tar.gz (3.5 MB view details)

Uploaded Aug 20, 2026 Source

Built Distribution

If you're not sure about the file name format, learn more about wheel file names.

The dropdown lists show the available interpreters, ABIs, and platforms. Enable javascript to be able to filter the list of wheel files.

hawk-2.4.0-py3-none-any.whl (792.4 kB view details)

Uploaded Aug 20, 2026 Python 3

File details

Details for the file hawk-2.4.0.tar.gz.

File metadata

Download URL: hawk-2.4.0.tar.gz
Upload date: Aug 20, 2026
Size: 3.5 MB
Tags: Source
Uploaded using Trusted Publishing? Yes
Uploaded via: twine/7.0.0 CPython/3.13.14

File hashes

Hashes for hawk-2.4.0.tar.gz
Algorithm	Hash digest
SHA256	`998f05f687ac01d90067d9cbaaf7118258af44860b4ac784bb3fe07c4c6ee7ee`
MD5	`772fddcf4ce43febbc1b153c59a9f9ab`
BLAKE2b-256	`5438658d53a5e59a3ac5ab6ff8d93df5a97c3a2774df53fa28fc7548894e1a74`

See more details on using hashes here.

Provenance

The following attestation bundles were made for hawk-2.4.0.tar.gz:

Publisher: publish-pypi.yml on METR/hawk

Attestations: Values shown here reflect the state when the release was signed and may no longer be current.

Statement:
- Statement type: https://in-toto.io/Statement/v1
- Predicate type: https://docs.pypi.org/attestations/publish/v1
- Subject name: hawk-2.4.0.tar.gz
- Subject digest: 998f05f687ac01d90067d9cbaaf7118258af44860b4ac784bb3fe07c4c6ee7ee
- Sigstore transparency entry: 2539212904
- Sigstore integration time: Aug 20, 2026
Source repository:
- Permalink: METR/hawk@b44eca56bd04d0f6f61faf6f90cd331f3a8aefbd
- Branch / Tag: refs/tags/v2.4.0
- Owner: https://github.com/METR
- Access: public
Publication detail:
- Token Issuer: https://token.actions.githubusercontent.com
- Runner Environment: github-hosted
- Publication workflow: publish-pypi.yml@b44eca56bd04d0f6f61faf6f90cd331f3a8aefbd
- Trigger Event: release