@jive-ai/cli
v0.0.47
Published
The official CLI for Jive AI - autonomous development workflows powered by Claude.
Readme
Jive CLI
The official CLI for Jive AI - autonomous development workflows powered by Claude.
Installation
npm
npm install -g @jive-ai/cliDocker
docker pull jiveai/task:latestConfiguration
Environment Variables
The CLI can be configured using environment variables:
JIVE_TASK_IDLE_TIMEOUT- Minutes of idle time before a task automatically terminates (default: 5). Tasks are ephemeral and will gracefully shutdown after this period of inactivity, freeing up runner capacity. Session data is preserved for resume.- Development:
JIVE_TASK_IDLE_TIMEOUT=10(10 minutes - more forgiving) - Production:
JIVE_TASK_IDLE_TIMEOUT=5(5 minutes - default) - CI/CD:
JIVE_TASK_IDLE_TIMEOUT=30(30 minutes - long builds)
- Development:
JIVE_API_KEY- Your Jive API authentication key (required for task runners)JIVE_TEAM_ID- The team ID to run tasks for (required for task runners)JIVE_API_URL- API endpoint (defaults to production)JIVE_WS_URL- WebSocket endpoint (defaults to production)JIVE_LOG_FILE- MCP server log path (defaults to/tmp/jive-mcp.log)
Docker Image
The docker image is used for each task process container. Whenever you spin up a new task in Jive, it will start a new container with this image. You can specify your own custom image in the project settings as long as the same entrypoint script is provided. The task runner relies on it.
Building Your Own Images
You can extend the Jive task runner image to add custom tools or dependencies:
FROM jiveai/task:latest
# Add your custom dependencies
RUN apt-get update && apt-get install -y --no-install-recommends \
python3 \
python3-pip \
&& rm -rf /var/lib/apt/lists/*
# Install additional tools
RUN pip install --break-system-packages black ruffNote: The base image is Debian-based (bookworm-slim), so use apt-get for package management.
Publishing the Docker Image (Maintainers)
One-Time Setup
Login to Docker Hub:
docker loginCreate a buildx builder for multi-platform builds:
docker buildx create --name mybuilder --use docker buildx inspect --bootstrap
Publishing a New Version
After publishing a new version to npm, publish the corresponding Docker image:
cd cli
npm run docker:publishThis will:
- Build for both
linux/amd64andlinux/arm64platforms - Tag with both
latestand the current package version (e.g.,0.0.31) - Push both tags to Docker Hub
The image pulls the CLI from npm (using the @jive-ai/cli package), so make sure you've published to npm first.
Manual Steps
If you need more control:
# Build only (no push)
npm run docker:build:public
# Push to Docker Hub
npm run docker:pushDevelopment
Local Development with Docker
To test the Docker image with your local code changes:
npm run docker:build:localThis builds an image using the local CLI tarball instead of pulling from npm.
