zeekay
9e93ffbeb2
`import hanzo` ran the CLI's module body, so it needed click, rich, httpx and everything else in the list before it could tell you its own version — on a bare interpreter it did not import at all. It also read `__version__` back off this module from `cli`, a cycle that only worked because the assignment happened to sit above the import. Both go away by naming things instead of importing them. `__getattr__` (PEP 562) resolves a member on first touch and caches it, so `hanzo.cloud` costs nothing until somebody wants the cloud client, and the second touch is a dict lookup. A member that is not installed says which distribution to install rather than raising ModuleNotFoundError from somewhere inside us. This is the shape the npm package and the Rust crate already had: one front door, everything behind it, nothing loaded that was not asked for. One table maps a name to its distribution, so adding a member is one line.
hanzo
Python orchestration helpers for Hanzo AI, and the older Python implementation of the Hanzo CLI.
The hanzo command lives elsewhere now
The Hanzo CLI is a native binary. Install it with:
curl -fsSL https://hanzo.sh | sh
hanzo auth login
That gives you hanzo (and hanzo-node, a symlink to the same build) with one
command group per Hanzo Cloud product. This package's console script is
hanzo-py — deliberately not hanzo, so it cannot shadow the real CLI on
your PATH:
pip install hanzo
hanzo-py --help
Nothing here is being removed. If you have scripts on hanzo-py, they keep
working; new work should target the native CLI.
What's still useful here
The Python library pieces, importable without touching the CLI:
from hanzo.batch_orchestrator import BatchOrchestrator
orchestrator = BatchOrchestrator()
results = await orchestrator.run_batch([
"Summarize the incident report",
"Draft the follow-up email",
])
from hanzo.memory_manager import MemoryManager
memory = MemoryManager()
memory.add_to_context("user", "What changed in the last deploy?")
context = memory.get_context()
from hanzo.fallback_handler import FallbackHandler
handler = FallbackHandler()
result = await handler.handle_with_fallback(
primary_fn=api_call,
fallback_fn=local_inference,
)
For the typed Hanzo Cloud API client, install hanzoai.
Configuration
HANZO_API_KEY=sk-...
HANZO_BASE_URL=https://api.hanzo.ai
Model ids come from the cloud catalog, not from this file — zen5,
zen5-coder, zen5-flash, zen5-mini, zen5-pro, enso, enso-flash,
enso-ultra. Ask the cloud rather than hardcoding: hanzo models list.
Development
cd pkg/hanzo
uv sync --all-extras
uv build
License
Apache 2.0