Crate cognis 

cognis

Implementation layer for the Cognis LLM framework. This crate provides concrete chat model integrations, agent execution, chains, memory strategies, document loaders, text splitters, embedding providers, and built-in tools.

Chat Model Providers

Each provider is gated behind a feature flag:

Feature	Provider
anthropic	Anthropic Claude
openai	OpenAI GPT
google	Google Gemini
ollama	Ollama (local)
azure	Azure OpenAI
all-providers	All of the above

Quick Example

use cognis::chat_models::anthropic::ChatAnthropic;
use cognis_core::runnables::Runnable;
use serde_json::json;

let model = ChatAnthropic::new("claude-sonnet-4-20250514");
let result = model.invoke(json!({"messages": []}), None).await.unwrap();

Modules

• chat_models – Chat model implementations for each provider.
• embeddings – OpenAI and Ollama embedding providers.
• agents – Agent executor with a pluggable middleware pipeline.
• chains – LLM chain, conversation chain, and sequential chain.
• memory – Buffer, window, and summary memory strategies.
• document_loaders – Text, CSV, JSON, and directory document loaders.
• text_splitter – Character, recursive, markdown, HTML, JSON, code, and token splitters.
• tools – Calculator, shell, and JSON query tools.

Re-exports

pub use cognis_core as core;

Modules

agents

Agent module providing middleware, tool-calling agents, structured output support, and output parsers for converting raw LLM text into structured agent actions.

cache

LLM response caching backends.

caching

API response caching for LLM calls.

callbacks

Centralized callback system for the entire execution lifecycle.

chains

Chain abstractions for composing prompts, models, and sequential pipelines.

chat_models

Chat model implementations, wrappers, and provider registry.

chat_sessions

Chat session manager with persistence and lifecycle management.

document_loaders

Document loader implementations for ingesting data from various sources.

document_transformers

Document transformer pipeline for processing, filtering, and enriching documents.

embeddings

Embeddings factory and provider registry.

evaluation

Evaluation framework for LLM outputs.

indexing

Indexing pipeline for incremental document ingestion.

memory

Conversation memory systems for managing chat history in chains.

output_parsers

Output parsers with LLM-based error correction and structured extraction.

prompts

Higher-level prompt management for Cognis.

providers

Provider integration framework for connecting to LLM APIs.

resilience

Resilience patterns for LLM API calls.

retrievers

Retriever implementations that compose and extend BaseRetriever.

stores

Key-value store implementations.

streaming

Streaming utilities for LLM responses.

text_splitter

text_splitters

Text splitting strategies for document chunking.

tools

Concrete tool implementations for use with the agent executor.

vectorstores