Operative
Computer use agent powered by Claude. Provides a Docker container environment where Claude interacts with a virtual desktop via screenshot, mouse, and keyboard tools, with a Streamlit UI for human interaction.
Image: ghcr.io/hanzoai/operative:latest
Repo: github.com/hanzoai/operative
Note: Computer use is a beta feature. It poses unique risks distinct from standard API features. To minimize risks:
- Use a dedicated virtual machine or container with minimal privileges.
- Avoid giving the model access to sensitive data such as account login information.
- Limit internet access to an allowlist of domains.
- Require human confirmation for decisions with real-world consequences.
Instructions found in web content or images may override user instructions. Isolate the agent from sensitive data and actions.
What It Does
- Runs a full Linux desktop (X11/Xvfb + VNC) inside Docker
- Claude controls the desktop via screenshot/mouse/keyboard tool calls
- Supports Anthropic API (direct), AWS Bedrock, and Google Vertex
- Streamlit UI for sending instructions and watching the agent work
- Built-in bash and file editing tools alongside computer control
Quick Start
Anthropic API
export ANTHROPIC_API_KEY=%your_api_key%
docker run \
-e ANTHROPIC_API_KEY=$ANTHROPIC_API_KEY \
-v $HOME/.anthropic:/home/operative/.anthropic \
-p 5900:5900 \
-p 8501:8501 \
-p 6080:6080 \
-p 8080:8080 \
-it ghcr.io/hanzoai/operative:latest
Bedrock
export AWS_PROFILE=<your_aws_profile>
docker run \
-e API_PROVIDER=bedrock \
-e AWS_PROFILE=$AWS_PROFILE \
-e AWS_REGION=us-west-2 \
-v $HOME/.aws:/home/operative/.aws \
-v $HOME/.anthropic:/home/operative/.anthropic \
-p 5900:5900 -p 8501:8501 -p 6080:6080 -p 8080:8080 \
-it ghcr.io/hanzoai/operative:latest
Vertex
gcloud auth application-default login
export VERTEX_REGION=%your_vertex_region%
export VERTEX_PROJECT_ID=%your_vertex_project_id%
docker run \
-e API_PROVIDER=vertex \
-e CLOUD_ML_REGION=$VERTEX_REGION \
-e ANTHROPIC_VERTEX_PROJECT_ID=$VERTEX_PROJECT_ID \
-v $HOME/.config/gcloud/application_default_credentials.json:/home/operative/.config/gcloud/application_default_credentials.json \
-p 5900:5900 -p 8501:8501 -p 6080:6080 -p 8080:8080 \
-it ghcr.io/hanzoai/operative:latest
Accessing the Interface
Once the container is running, open http://localhost:8080 for the combined agent chat + desktop view.
The container stores settings (API key, custom system prompt) in ~/.anthropic/. Mount this directory to persist settings between runs.
| Port | Service |
|---|---|
| 8080 | Combined interface |
| 8501 | Streamlit UI only |
| 6080 | noVNC desktop only |
| 5900 | VNC (for VNC clients) |
Screen Size
Set resolution with WIDTH and HEIGHT environment variables:
docker run \
-e ANTHROPIC_API_KEY=$ANTHROPIC_API_KEY \
-e WIDTH=1920 -e HEIGHT=1080 \
-v $HOME/.anthropic:/home/operative/.anthropic \
-p 5900:5900 -p 8501:8501 -p 6080:6080 -p 8080:8080 \
-it ghcr.io/hanzoai/operative:latest
XGA resolution (1024x768) is recommended. For higher resolutions, scale screenshots down to XGA and map coordinates back proportionally. For smaller displays, add black padding to reach 1024x768.
Development
./setup.sh # configure venv, install dev dependencies, pre-commit hooks
docker build . -t operative:local
export ANTHROPIC_API_KEY=%your_api_key%
docker run \
-e ANTHROPIC_API_KEY=$ANTHROPIC_API_KEY \
-v $(pwd)/operative:/home/operative/ \
-v $HOME/.anthropic:/home/operative/.anthropic \
-p 5900:5900 -p 8501:8501 -p 6080:6080 -p 8080:8080 \
-it operative:local
The docker run command mounts the repo inside the container with auto-reloading enabled.
Credits
Based on Anthropic's computer use demo. See upstream repository for original implementation.
License
MIT
Metadata
Release files for hanzo-operative 0.1.1
For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.
Source distribution (sdist)
| File | Size | Uploaded | Provenance | Actions |
|---|---|---|---|---|
| hanzo_operative-0.1.1.tar.gz | 24.2 kB | Without provenance | Details for hanzo_operative-0.1.1.tar.gz |
Built distribution (wheel)
| File | Interpreter | ABI | Platform | Actions Reset |
|---|---|---|---|---|
| hanzo_operative-0.1.1-py3-none-any.whl | Python 3 | none | any | Details for hanzo_operative-0.1.1-py3-none-any.whl |
Total release size: 52.0 kB
Release files / hanzo_operative-0.1.1.tar.gz
| Download URL | hanzo_operative-0.1.1.tar.gz |
|---|---|
| Size | 24.2 kB |
| Tags | Source |
|
SHA-256 checksum How to use checksums |
8f4e919713197f8d8247c292e156d694718dce8e983c8b85bbedc217edb42728
|
|
BLAKE2b-256 checksum How to use checksums |
894cae4468f6d909a66bd05f92a7ee8921e2d1f1311127dba2ff54e1cedba04b
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
twine/7.0.0 CPython/3.14.7
|
Release files / hanzo_operative-0.1.1-py3-none-any.whl
| Download URL | hanzo_operative-0.1.1-py3-none-any.whl |
|---|---|
| Size | 27.8 kB |
| Tags | Python 3 |
|
SHA-256 checksum How to use checksums |
1fb32505e847f5fa518cc6b42c2d8c4ffb1a0e33334574041d142101469ffec0
|
|
BLAKE2b-256 checksum How to use checksums |
dbf1360f8a360b41a123876e3b8d28731533c26f717d82c452ffe768f3d12833
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
twine/7.0.0 CPython/3.14.7
|
Security for hanzo-operative 0.1.1
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
No provenance found
None of the files in this release were uploaded with provenance metadata.
A note on package security
PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.
To learn more about protecting yourself from malicious actors, visit our package security documentation.