[New Env] Add web search env with Google Search (via Serper api)

[Lawhy]

Add a new environment that supports web search.

[burtenshaw]

Hi @Lawhy, Nice env! It looks like your env does not conform to the standard structure. So we need to add a standard structure.

The easiest way is to use the openenv cli to initialize a new project, and then copy in your cod to that skeleton. You can do this like so:

• Merge main
• Reinstall uv sync
• cd src/envs
• openenv init websearch_env_new
• copy your business logic into the new skeleton
• mv websearch_env_new websearch_env
• push your changes.

Once you're done this, your env will be compatible with the hub and have project matter to be installed and run. You can get help along the way from openenv --help.

[Lawhy]

Hi @burtenshaw, thanks for quick responses! I have merged mainline and trying to use openenv init. One thing I'm confused is that I can see openenv_core is used through this command e.g.,:

from openenv_core.client_types import StepResult
from openenv_core.env_server.types import State
from openenv_core.http_env_client import HTTPEnvClient

but it is not installed with uv sync

[burtenshaw]

One thing I'm confused is that I can see openenv_core is used through this command e.g.,: but it is not installed with uv sync

openenv-core should be installed via the env's toml. It's in the template here. The pip package might not be upto dat. So try using this instead "openenv-core @ git+https://github.com/meta-pytorch/OpenEnv.git@main#subdirectory=src/core" .

Feel free to push your updated project and I can try it out.

[Lawhy]

Thanks @burtenshaw, I have pushed a new version

[Lawhy]

Seems that openenv init doesn't create pyproject.toml, requirements.txt and some other files shown in README.md; I have rebased the branch to the latest main.

[burtenshaw]

Seems that openenv init doesn't create pyproject.toml, requirements.txt and some other files shown in README.md; I have rebased the branch to the latest main.

Strange. Are you doing openenv init <a_new_directory_name> ?

[burtenshaw]

@Lawhy I took a look and adapted your env to the new skeleton: Lawhy#1 I think you'll just need to resolve dependencies and test.

Sorry about this, it's just challenging because you started before the init command existed.

[Lawhy]

Thanks @burtenshaw, I actually did openenv init and I installed the package by uv pip install -e OpenEnv locally. Not sure why the cli is not up-to-date.

[Lawhy]

Hi @burtenshaw, is there anything I need to check? My local test has no problem.

[burtenshaw]

Hi @burtenshaw, is there anything I need to check? My local test has no problem.

Yeah, it runs but I just wanted to make sure you've tested it functionally. Could you push it to hugging face hub with openenv push when it's ready? And link here.

[github-actions]

✅ Validation succeeded for websearch_env

Your env passes the vibe check. However, most environments should go straight to the hub, they will automatically be added to the official Env Hub collection on a nightly basis. Environments in the official specification repo are only meant to demonstrate usage of a specific spec feature for educational purposes. Re-run locally with:

• openenv validate --verbose src/envs/websearch_env

[OK] websearch: Ready for multi-mode deployment

Supported deployment modes:
  [YES] docker
  [YES] openenv_serve
  [YES] uv_run
  [YES] python_module

Usage examples:
  cd websearch_env && uv run server
  cd websearch_env && openenv build
  cd websearch_env && openenv push

You can deploy the environment to Hugging Face Spaces by running openenv push.

• Success run: https://github.com/meta-pytorch/OpenEnv/actions/runs/19346757979

[Lawhy]

Hi @burtenshaw, it's working now: https://huggingface.co/spaces/lawhy/web_search; it will require API key from https://serper.dev/ or set it up in Hugging Face secret.