brainwires-prompting 0.11.0

Adaptive prompting techniques, K-means task clustering, and temperature optimization for the Brainwires Agent Framework
Documentation

brainwires-prompting

Crates.io Documentation License

Adaptive prompting techniques, K-means task clustering, and temperature optimization for Brainwires agents.

What lives here

  • library::TechniqueLibrary + techniques::* — 15-technique library from the adaptive-selection paper (chain-of-thought, self-consistency, tree-of-thought, …).
  • clustering::TaskCluster + TaskClusterManager — K-means task clustering by semantic-vector similarity (linfa-backed).
  • generator::PromptGenerator — dynamic prompt generation. With the knowledge feature (default-on) integrates BKS (Behavioral Knowledge System) / PKS (Personal Knowledge System) / SEAL feedback to adapt outputs over time.
  • learning::PromptingLearningCoordinator — technique-effectiveness tracking + BKS promotion logic.
  • temperature::TemperatureOptimizer — adaptive temperature optimisation per task cluster.
  • seal::SealProcessingResult — feedback hook used by generator.
  • storage::ClusterStorage (gated by storage feature) — SQLite-backed cluster persistence.

Features

Feature Default Notes
knowledge yes Pulls brainwires-knowledge for BKS/PKS-aware prompt generation
storage no SQLite cluster store (rusqlite)

Most modules (generator, learning, temperature) reference BKS/PKS unconditionally, hence the knowledge default. The standalone bits (clustering, library, techniques, seal) work without it.

Usage

[dependencies]
brainwires-prompting = "0.11"

use brainwires_prompting::{TechniqueLibrary, PromptingTechnique};

let library = TechniqueLibrary::default();
let technique = library.lookup(PromptingTechnique::ChainOfThought);

See also

License

MIT OR Apache-2.0