proserpina 0.1.0

Multi-agent critique and cross-examination pipeline for documents requiring intellectual rigor — provider-agnostic interaction-graph engine with pluggable LLM backends
Documentation 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28

// Copyright (C) 2026 Industrial Algebra
// SPDX-License-Identifier: AGPL-3.0-only

//! Backends: concrete implementations of [`crate::agent::Agent`].
//!
//! - [`echo::EchoAgent`] is the deterministic, dependency-free reference
//!   backend used to drive the engine in tests.
//! - The HTTP backend ([`http`] module, behind the `backend-http` feature)
//!   calls an OpenAI-compatible chat-completions endpoint and is the first
//!   real LLM backend.
//! - The multi-provider roster ([`roster`] module, behind the `backend-http`
//!   feature) assigns authed providers to critic personas at random.
//! - The credentials config ([`credentials`] module, behind the `backend-http`
//!   feature) reads provider keys and model overrides from a TOML file so
//!   Proserpina can reach providers whose keys aren't in the environment.

pub mod echo;

#[cfg(feature = "backend-http")]
pub mod credentials;

#[cfg(feature = "backend-http")]
pub mod http;

#[cfg(feature = "backend-http")]
pub mod roster;

pub use echo::EchoAgent;