implicant 0.2.0

Python client SDK for the Implicant FHE inference platform

Implicant SDK

Python client for the Implicant FHE inference platform. Encrypts one row locally under BGV, sends the ciphertext to the platform for homomorphic evaluation, and decrypts the result. The server never sees plaintext; the secret key never leaves the process.

The classifier head runs server-side under FHE (Variant B): the platform returns one ciphertext with the class scores packed contiguously, which the client decrypts (unsigned) and decodes with decode_class_scores before threshold/argmax. W / b never reach the client.

Pure Python — all crypto comes from implicant-fhe (helut.client). Contracts: docs/architecture.md and platform/docs/FHE_INTERFACE_SPEC.md.

Install (developers)

implicant-fhe (the helut crypto wheel) is not on a public index — resolve it from the bundled ./wheels/. Wheels ship for cp310–cp313 (macOS-arm64 + manylinux-x86_64) only; Python 3.14 has no wheel, so use a 3.10–3.13 interpreter.

python3.13 -m venv .venv313
.venv313/bin/pip install -e ".[app,dev]" --find-links wheels/
.venv313/bin/pytest tests/ -v      # 79 pass in ~3.5 min — the real smoke test

Desktop app (end users)

Non-technical users install a double-click app instead of using Python:

• macOS (Apple chip): open the .dmg and drag Implicant to Applications.
• Linux (Intel/AMD chip): run the .AppImage.

Other platforms are not supported yet; the app says so on launch. Building the installers is documented in docs/INSTALL_SMOKE_TEST.md.

Use

from implicant import ImplicantClient
from implicant.transport import HttpxTransport

client = ImplicantClient(
    HttpxTransport(base_url="https://api.implicant.example", api_key="imp_..."),
    key_cache_dir="~/.implicant/keys",
)
result = client.predict(
    "breast-cancer",
    {"mean_radius": 14.2, "mean_texture": 20.1},
    class_names=("benign", "malignant"),
)
print(result.prediction.label)

implicant predict --model breast-cancer --input row.json

License

Apache-2.0.