A framework for production AI agents.
Documentation · Getting Started · CLI Reference · Community Slack
Zil composes with ADK, A2A, MCP, DeepEval, and OpenTelemetry to provide a declarative manifest format and CLI for building, validating, auditing, packaging, and deploying AI agents.
uv pip install zil-aiFor agent creation with ADK:
uv pip install 'zil-ai[adk]'# Scaffold a new agent project
zil init my-agent
cd my-agent && source .venv/bin/activate
# Validate the project
zil validate
# Security audit
zil audit --fix
# Run the agent interactively
zil run
# Or start the ADK web UI
zil web| Command | Description |
|---|---|
zil init |
Scaffold a new agent project |
zil validate |
Validate project against the manifest schema |
zil audit |
Agent-native security audit (injection, leakage, identity hardening) |
zil eval run |
Run evaluation suites |
zil eval generate |
LLM-powered eval case synthesis |
zil run |
Run the agent interactively |
zil web |
Start the ADK web UI for testing |
zil pack |
Build a versioned .zil archive |
zil push |
Push archives to an OCI-compatible registry |
zil deploy |
Deploy to Google Cloud Run with eval gating |
import zil
root_agent = zil.create_agent(
tools=[], # your tool functions
enable_guardrails=True, # runtime guardrail engine
enable_telemetry=True, # OpenTelemetry tracing
enable_memory=True, # long-term memory from adapters/memory.yaml
)See the SDK Reference for the full API.
Long-term memory is framework-neutral and works across ADK and OpenHands —
see docs/memory.md.
The examples/hello-agent directory contains a minimal reference agent you can run immediately:
cd examples/hello-agent
uv pip install -r requirements.txt
zil validate && zil audit && zil run- Slack — Join the Zil community for questions, feedback, and discussion
- GitHub Discussions — Ask questions and share ideas
- Issues — Report bugs or request features
We welcome contributions! See CONTRIBUTING.md for development setup, testing, and PR guidelines.
Built by FluentData