Reasoning Python Packages

pageindex

📑 PageIndex: Document Index for Vectorless, Reasoning-based RAG

agentmake

AgentMake AI: a kit for developing agentic AI applications that support 24 AI backends and and work with 7 agentic components, such as tools and agents. (Developer: Eliran Wong) Supported backends: anthropic, azure, azure_any, cohere, custom, deepseek, genai, github, github_any, googleai, groq, llamacpp, mistral, ollama, openai, vertexai, xai

miniful

A minimal fuzzy library

graqle

Give your AI tools architecture-aware reasoning. Knowledge graph from any codebase — 14 LLM backends, impact analysis, confidence scores. Works with Claude Code, Cursor, VS Code Copilot.

opendsstar

OpenDsStar is an open-source implementation of the DS-Star agent that replaces file-based workflows with a flexible, tool-centric architecture. It supports incremental execution, reuses intermediate results, and makes complex multi-step agents more modular, efficient, and extensible.

nocturnusai

Verified knowledge for AI agents. Compress context, extract and store facts, define rules, and ask questions — get deterministic answers with proof, not LLM guesses. Connect agents via MCP, Python SDK, TypeSc

ripple-down-rules

A python implementation of the various ripple down rules versions, including Single Classification (SCRDR), Multi Classification (MCRDR), and Generalised Ripple Down Rules (GRDR).

vectorless

Knowing by reasoning, not vectors. ⭐ Star this repo if you find it useful.

invinoveritas

▎ Lightning-native AI reasoning, decisions, memory, orchestration, agent marketplace, and message board. Agents earn sats, DM each other, and replicate autonomously. Pay-per-use via Bitcoin ▎ Lightning (L402 · Bearer · NWC). No subscriptions. No KYC.

awex

A high-performance RL training-inference weight synchronization framework, designed to enable second-level parameter updates from training to inference in RL workflows

ntloss

PyPI package for Number Token Loss (ICML 2025)

coconut-pytorch

Implementation of 🥥 Coconut, Chain of Continuous Thought, in Pytorch

sphereql

Spherical coordinate knowledge representation — fast semantic search, visualization, and analysis

hrm-pytorch

Exploration into the proposed architecture from Sapient Intelligence of Singapore 🇸🇬

cognigraph

Give your AI tools architecture-aware reasoning. Knowledge graph from any codebase — 14 LLM backends, impact analysis, confidence scores. Works with Claude Code, Cursor, VS Code Copilot.

cmem-plugin-reason

Perform reasoning tasks and validate OWL consistency.

entity-query-language

Provides a relational query language that is pythonic and intuitive. Entity Query Language serves as a front end to other query languages like sql or prolog

abstractllm

A unified interface for large language models with support for OpenAI, Anthropic, Hugging Face, Ollama, and MLX

prolog-reasoner

SWI-Prolog as a logic calculator for LLMs — MCP server and Python library

tensor-theorem-prover

Customizable first-order logic theorem prover supporting approximate vector similarity in unification

vulcan-core

AI-Hybrid Rules Engine for Logical Reasoning.

cairntir

Memory-first reasoning layer for Claude Code. Kills cross-chat AI amnesia. MCP server, verbatim storage, prediction-bound drawers, belief-as-distribution retrieval.

argus-debate-ai

Debate-native multi-agent AI framework that eliminates LLM hallucination, overconfidence, and opacity. Structures knowledge evaluation as adversarial multi-round debates over a Conceptual Debate Graph (C-DAG) with Bayesian aggregation, CRUX protocol, 8 novel Evolution extensions, 27+ LLM providers, 50+ tools, and W3C-compliant provenance tracking.

beyondbench

[ICLR 2026 Accepted paper] BeyondBench: Contamination-Resistant Evaluation of Reasoning in Language Models