fluctlightdb 0.4.3

Python client for FluctlightDB — brain-native memory for AI agents

fluctlightdb

Python client for FluctlightDB — a brain-native memory store for AI agents.

Install

On Debian/Ubuntu 23.04+, Debian 12+, and Fedora 38+, system Python is PEP 668 externally managed — bare pip install fails with externally-managed-environment. Use a virtual environment (same as any other PyPI library):

python3 -m venv .venv
source .venv/bin/activate   # Windows: .venv\Scripts\activate
pip install fluctlightdb

From a clone of this repo you can also run:

./scripts/install-python-client.sh

Optional in-process recall (Rust extension, when wheels are available for your platform):

pip install "fluctlightdb[native]"
# or: pip install fluctlightdb-native

No cargo or Rust toolchain required for the HTTP client.

Quick start (HTTP — like qdrant-client)

Run a FluctlightDB server (download a release binary or use your own deployment), then:

from fluctlightdb import FluctlightClient

client = FluctlightClient(
    base_url="http://127.0.0.1:8792",
    api_key="your-key",
)

client.experience("user prefers dark mode", context="settings")
print(client.activate_lite("theme preference"))

Or use environment variables:

export FLUCTLIGHT_SERVE_URL=http://127.0.0.1:8792
export FLUCTLIGHT_API_KEY=your-key

from fluctlightdb import FluctlightClient

client = FluctlightClient.from_env()
print(client.activate("dark mode"))

In-process brain (like sqlite3)

When fluctlightdb-native is installed:

from fluctlightdb import connect

brain = connect("/tmp/my-agent-brain")
brain.experience("user prefers dark mode", context="settings")
print(brain.activate("dark mode"))

Read-only recall path: get_recall_client(path).

In-process recall (read-only helper)

Docs

• Getting started
• HTTP API (OpenAPI)

License

MIT — see LICENSE.