Install Kayak

Install Kayak in the same environment as a usable mojo CLI. That is the real requirement. UV, pip, and Pixi can all work — the deciding factor is whether Kayak can actually discover a usable mojo command when your Python code runs.

Setup shape	Result
`uv add kayak` or `pip install kayak` with no usable `mojo` CLI	`numpy_reference` only
`uv add kayak` in an environment where `mojo` is already installed and discoverable	`mojo_exact_cpu` is available
`pixi add python=3.11 ... mojo` plus `pixi add --pypi kayak`	`mojo_exact_cpu` is available

Choose The Install Path That Matches Your Environment

I want Mojo nowI already have MojoI only need Python first

Use one environment where Python, Mojo, and Kayak live together.

pixi add python=3.11 "mojo>=0.26.3.0.dev2026041020,<0.27"
pixi add --pypi kayak

Then verify:

pixi run python - <<'PY'
import kayak
print(kayak.available_backends())
print(kayak.backend_info(kayak.MOJO_EXACT_CPU_BACKEND).availability_reason)
print(kayak.mojo_bridge_info())
PY

If a usable mojo CLI is already in the active environment or on PATH, you can keep using UV or pip.

mojo --version
uv add kayak
uv run python - <<'PY'
import kayak
print(kayak.available_backends())
print(kayak.backend_info(kayak.MOJO_EXACT_CPU_BACKEND).availability_reason)
print(kayak.mojo_bridge_info())
PY

Install Kayak by itself:

uv add kayak

In that setup Kayak still works, but only the NumPy backend is available:

('numpy_reference',)

Recommended Verification

Run this after install regardless of environment manager:

python - <<'PY'
import kayak

print(kayak.available_backends())
print(kayak.backend_info(kayak.MOJO_EXACT_CPU_BACKEND).availability_reason)
print(kayak.mojo_bridge_info())
print(kayak.doctor())
PY

Use this to verify four facts:

the package imports correctly
the exact backend is visible
Kayak can explain how it found Mojo
the public diagnostics report matches the environment you intended

One Verified Mojo-Capable Path

Kayak currently validates the Mojo dependency in the range:

>=0.26.3.0.dev2026041020,<0.27

One verified way to satisfy that is a Pixi workspace that already includes Modular's max-nightly channel and conda-forge:

pixi add python=3.11 "mojo>=0.26.3.0.dev2026041020,<0.27"
pixi add --pypi kayak

Expected verification shape:

('numpy_reference', 'mojo_exact_cpu')
True
'Kayak can invoke Mojo via: ...'

Pixi is not required. It is just one clear way to keep Python, Mojo, and Kayak in the same environment.

UV Plus A Preinstalled Mojo CLI

If the environment already owns a working mojo CLI, UV is also a correct Mojo-capable setup:

mojo --version
uv add kayak
uv run python - <<'PY'
import kayak
print(kayak.available_backends())
print(kayak.backend_info(kayak.MOJO_EXACT_CPU_BACKEND).availability_reason)
print(kayak.mojo_bridge_info())
PY

This is the right path when your environment management is already settled and you only need Kayak to bind to that existing Mojo install.

Optional Store Adapters

The core SDK does not require external database clients. Install only the adapter you actually plan to use.

Adapter	Install
LanceDB	`uv add lancedb pyarrow` or `pixi add --pypi lancedb pyarrow`
Qdrant	`uv add qdrant-client` or `pixi add --pypi qdrant-client`
Weaviate	`uv add weaviate-client` or `pixi add --pypi weaviate-client`
Chroma	`uv add chromadb` or `pixi add --pypi chromadb`
PgVector	`uv add "psycopg[binary]" pgvector` or `pixi add --pypi "psycopg[binary]" pgvector`

What Not To Assume

If your intent is "I want the Mojo backend," do not stop at:

uv add kayak

That installs the Python SDK, not Mojo itself. In that environment available_backends() will normally return only:

('numpy_reference',)

The SDK works, but the exact CPU Mojo backend is not available yet.

How Kayak Finds Mojo

Kayak discovers Mojo in this order:

KAYAK_MOJO_CLI
a usable mojo binary in the active Python environment
mojo on PATH
pixi run mojo

If you have multiple Mojo installs, set KAYAK_MOJO_CLI explicitly:

export KAYAK_MOJO_CLI=/full/path/to/mojo