add analytics guide#184
Conversation
✅ Deploy Preview for agent-sandbox ready!
To edit notification comments on pull requests, go to your Netlify project configuration. |
k8s-ci-robot
added
the
needs-ok-to-test
|
Hi @drogovozDP. Thanks for your PR. I'm waiting for a github.com member to verify that this patch is reasonable to test. If it is, they should reply with Once the patch is verified, the new status will be reflected by the I understand the commands that are listed here. DetailsInstructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes-sigs/prow repository. |
k8s-ci-robot
added
size/XXL
|
/ok-to-test |
k8s-ci-robot
added
ok-to-test
| fsGroup: 2000 | ||
| containers: | ||
| - name: python-sandbox | ||
| image: us-docker.pkg.dev/<PROJECT_ID>/analytics/analytics-tool:1.0.0 |
There was a problem hiding this comment.
Use the bash format please.
us-docker.pkg.dev/$PROJECT_ID/analytics/analytics-tool:1.0.0
| --description="Analytics Repo" | ||
| ``` | ||
|
|
||
| And now we can create our analytics agent-sandbox tool: |
There was a problem hiding this comment.
Prevent using the words: And now, our. Instead structure it as bullet points.
- Create a repository in Artifact Registry.
- Create analytics agent-sandbox.
- Deploy the changes.
There was a problem hiding this comment.
Paraphrased all "we", "our", and "you" moments
| print(f"<IMG>{img_str}</IMG>") | ||
| ``` | ||
|
|
||
| As you can see, in the end the code prints an encoded image. Inside the tool definition we use the regex expression to extract this string, decode, and plot it. |
There was a problem hiding this comment.
Can we add a sample output of how it will look like?
| """ | ||
| try: | ||
| # Split the command string into a list to safely pass to subprocess | ||
| args = shlex.split(request.command) |
There was a problem hiding this comment.
Validation to check malformed commands?
| args, | ||
| capture_output=True, | ||
| text=True, | ||
| cwd="/app" |
There was a problem hiding this comment.
Consider adding a timeout.
|
|
||
| ## Deploy analytics tools | ||
|
|
||
| This section describes how to build Docker image that defines analytics tool for an ADK agent, and push the Docker image to a Artifact Registry repository and deploy from it. |
There was a problem hiding this comment.
Nit:
This section describes how to build Docker image that defines analytics tool for an ADK agent, push the Docker image to a Artifact Registry repository and deploy the pushed image.
| kubectl port-forward "pod/jupyterlab-sandbox" 8888:8888 | ||
| ``` | ||
|
|
||
| Follow the `welcome.ipynb` notebook. |
There was a problem hiding this comment.
Can you reference this from the yaml file?
|
Can we add a PR description which explains the motivation for this? I am working on a tangent PR that allows stateful execution via jupyter client: #296. |
Description updated. |
| @@ -0,0 +1,110 @@ | |||
| # AI Analytics with agent-sandbox | |||
There was a problem hiding this comment.
Add a small overview of what this example does before going into the implementation details.
|
|
||
| ## Deploy analytics tools | ||
|
|
||
| This section describes how to build Docker image that defines analytics tool for an ADK agent, push the Docker image to a Artifact Registry repository and deploy the pushed image. |
There was a problem hiding this comment.
There is no ADK agent in this example ??
|
/lgtm |
k8s-ci-robot
added
the
lgtm
|
[APPROVALNOTIFIER] This PR is APPROVED This pull-request has been approved by: barney-s, drogovozDP The full list of commands accepted by this bot can be found here. The pull request process is described here DetailsNeeds approval from an approver in each of these files:
Approvers can indicate their approval by writing |
k8s-ci-robot
added
the
approved
7 participants
This example shows how to use agent-sandbox as a remote code execution backend for an LLM analytics agent. The agent runs in JupyterLab, defines a tool that sends Python code to a sandbox pod for execution, and renders the output (including charts) back in the notebook. The motivation is to give users a concrete end-to-end example of delegating unsafe code execution to an isolated sandbox.