One directory, three calls

Install

pip install "fluctlightdb[native]"

The [native] extra pulls the compiled Rust core. Without it you get the pure-Python fallback, which is slower and is not what the measurements were taken on. Use a virtual environment; do not install into the system Python.

A brain, in a few lines

This is the canonical sample. It was run against the published package, and it is the same sample the rest of the site shows. Do not treat a version pin you saw somewhere else as part of it.

from fluctlightdb import connect_embedded

brain = connect_embedded("./my-agent-brain")   # one directory per agent, in-process
brain.experience("User prefers dark mode", context="settings", salience=0.7)
brain.experience("Wallet balance is $42.10", context="ledger:wallet",
                 verified=True, source_uri="file://wallet.json")

hits = brain.activate("dark mode")             # cue -> ranked recalls
brain.checkpoint()                             # durable; survives restart

experience() writes a memory. The second call is verified and names the file it came from, which is what provenance means on this engine. activate("dark mode") recalls from a cue. checkpoint() is what makes the directory survive the process.

Offline, recall is lexical: a cue has to share tokens with what was stored. Paraphrase recall needs a semantic vector from an embedder you bring. Source: docs/EMBEDDINGS.md:9,24. Opens the repository in a new tab.

Where to go next

• Install, if you want only the pip line, the canonical sample and the checkpoint note.
• The documentation, including the return shapes and the mistakes agents make against this API.
• A glossary of engram, provenance, cue recall and consolidation.
• The playground, which talks to one embedded engine in a Python process.
• The preprint, if you want the argument and a citation.