PyPI agents

LLM Firewall: a safety layer between your application and any LLM API — input/output firewalling, PII redaction, jailbreak detection, and domain-specific guardrails.

A lightweight Python decorator for caching, rate limiting, and retrying LLM API calls.

A minimal, ergonomic Python library for interacting with LLMs

LLMIntent suite: semantic extraction, model registry, isolates, motifs, trajectories, IV layer-causal, and latent thought inspection

Local, folder-first LLM knowledge bases: format spec, CLI, MCP server, and agent skills.

Python library to support LLM testing.

LLM-based ranking and reasoning algorithms for search and recommendation

Intelligent routing, load balancing, and failover for LLM providers

LLM Runner Control is a CLI tool to interact with local and remote LLM, define multiple scenario variations (model × prompt × options), run it over a dataset and capture structured results for comparison or to build knowledge through meta analysis.

Paquete que permite trabajar con los Grandes Modelos de Lenguaje (LLM, por sus siglas en inglés) y los Grafos de Conocimiento (KG, por sus siglas en inglés)

Mirror official docs (via llms.txt) as plain Markdown for LLM ingestion and RAG.

Generate and validate llms.txt files — CLI and Python library, zero dependencies.

Unified interface to call OpenAI and Anthropic-compatible LLM APIs from Python

Python SDK for LLMSafeSpaces API

Zero-dependency, cross-platform LLM/GEO/AEO checker CLI. 21 checks, pure Python stdlib, no runtime dependencies.

LLM Surgery

A production-ready routing, fallback, and rate-limiting client for OpenAI-compatible LLM endpoints.

Find, download (over BitTorrent + webseeds), and SHA-256-verify LLM models from the LLM Torrents catalog.

Profile LLM gateway logs for prefix-cache reuse potential, and export anonymized replayable traces

Identifiers designed to survive being read, copied and re-emitted by language models

Passive LLM endpoint identity and stability instrumentation

Serve existing Markdown, Obsidian-style, or LLMWiki folders as cited agent-readable context.

MCP server providing agent capabilities for LM Studio: file I/O, terminal execution, web search, persistent memory, and skill loading.

Shared AWS Bedrock internals (SigV4 signing, event-stream decoding, Anthropic pricing) for lmux

A local-first AI orchestration framework

CLI tool to manage local LLM models and role-based server lifecycle

RAG Knowledge MCP Server — FastMCP + LanceDB + Hybrid Search

Localstorage for python with more tools

Use the OpenAI Python SDK with your own ChatGPT subscription

Deterministic package-map inference engine for Logion.

LoomEye HPC Agent — connect your cluster to LoomEye

loopy-agent: 19 Essential AI Concepts in one toolkit — agentic loops, gateway, guardrails, evals, caching, observability, MCP, multi-agent, middleware, plugins, state, safety, cost, drift, skills, verification, audit, streaming, and compliance.

LoRA adapter bridge between Unsloth and Ollama for dynamic adapter loading.

Python SDK for Lore — coordination memory for multi-agent AI systems: create runs, write events, and fetch read-your-writes context packs.

Pure Python ctypes wrapper for the Luxand LLM SDK

Recognizable alias for m3-memory's LangChain / LangGraph integration (drop-in Mem0 replacement, M3Store, M3Saver checkpointer, retriever, LCEL). Re-exports m3_memory.langchain.

A lightweight multi-agent coordination ledger for task handoff, context handoff, and quality gates.

Launch each project's AI coding agent in its own terminal, auto-tiled across every monitor

AIGP governance adapter for anthropic

AIGP governance adapter for Microsoft AutoGen multi-agent conversations

AIGP governance adapter for CrewAI multi-agent orchestration

AIGP governance adapter for llamaindex

AIGP governance adapter for openai

LLM-powered static analysis and repository understanding tool

Utility tool for converting various files to Markdown

Maton Agent Toolkit

The self-improving AI agent — creates skills from experience, improves them during use, and runs anywhere

Model agents and DAG-based multi-agent collaboration components for MatterLoop

Synap memory integration for Smolagents — memory tools and a turn recorder

Synap memory integration for Strands Agents — long-term memory, short-term context, and real-time anticipation

MCP server to validate agent health under AOS discipline — catch agents that silently fail in production.

Observability for AI agent loops: trace trees, metrics, loop detection, cost tracking. Zero dependencies.

An MCP which allows you to run commands in a terminal and get the output back.

MCP-Bastion security helpers for Microsoft AutoGen / Agent Framework tool messages.

MCP-Bastion security helpers for LangGraph nodes and tool calls.

MCP-Bastion security middleware for LiteLLM (100+ LLM providers via one OpenAI-compatible API).

MCP-Bastion security middleware for Ollama local models (OpenAI-compatible API).

MCP-Bastion security helpers for OpenAI Agents SDK tools and messages.

MCP-Bastion security helpers for Pydantic AI agent prompts and tool args.

MCP server for Mistral API with queueing and rate limiting

Expose any MCP server's tools as OpenAI function-calling tools, and execute OpenAI tool calls back against it.

Microsoft AutoGen agents testing helpers for MCP Test Harness

LangGraph agent graphs testing helpers for MCP Test Harness

LiteLLM proxy / router testing helpers for MCP Test Harness

Ollama local models testing helpers for MCP Test Harness

OpenAI Agents SDK testing helpers for MCP Test Harness

Pydantic AI agents testing helpers for MCP Test Harness

vLLM native KV offloading secondary tier backed by the MemKV context memory store (TieringOffloadingSpec).

Durable Anthropic Messages adapter for the MemoFlow Python SDK

Durable OpenAI Responses adapter for the MemoFlow Python SDK

Poison-resistant, Byzantine-fault-tolerant shared memory for Microsoft AutoGen agent teams — implements the autogen_core Memory protocol; one rogue agent can't corrupt the team's recall.

Poison-resistant, Byzantine-fault-tolerant shared memory + robust voting tools for CrewAI crews — one rogue agent can't corrupt what the crew remembers or how it votes.

Poison-resistant, Byzantine-fault-tolerant shared chat memory for LangChain agent swarms — one prompt-injected agent can't corrupt the shared history.

Bit-exact Byzantine-fault-tolerant state layer for federated learning and edge/defence swarms. Three layers — OR-Set CRDT (L1), an epistemic drift + human-in-the-loop layer (L2), and ACFA Q16.16 multi-Krum + G-Set (L3) — keep heterogeneous nodes byte-identical; one prompt-injected or equivocating node can't fork, poison, or move the agreed state. Every write is signed, attributable, and exactly replayable. Architected for large swarms; bit-exact by construction.

Mengram memory tools for CrewAI agents — semantic, episodic, and procedural memory (workflows that learn from failures).

A polished Python chat workspace for Ollama and OpenRouter models

A from-scratch grounded RAG engine over biomedical literature: every ML component trained in-house, every answer cited, verified, or refused.

Framework for benchmarking, evaluating, and orchestrating Retrieval-Augmented Generation pipelines.

Deterministic microscopy image analysis for AI agents, over MCP. Point your Claude/Cursor/Codex agent at an image and it segments, counts, and measures cells — Fiji/ImageJ-equivalent, every number traceable to a tool, never guessed.

Entra ID Agent Container (sidecar) authentication library for Microsoft Agents

Integration library for Microsoft Agents with Teams

The LLM stack you can read and understand: Gemma 4 and Llama on CUDA, built from explicit C++23 components.

A developer-facing memory layer for AI agents (MCP server).

Evolutionary RAG system with holographic memory compression

A dynamic, need-based multi-agent AI framework with session memory, action tracking, tool approval, and browser automation

A minimal coding agent

General purpose agent harness inspired by Claude Code

A compact transactional runtime for verified coding agents.

A lean, snappy fullscreen terminal client for Claude Code

A minimal terminal AI agent inspired by minimal-agent.com.

Local first personal AI agent powered by the Mirai cloud API

MkDocs plugin that submits your docs site to AgentReady after every build, making it instantly queryable by AI agents via MCP.

Closed-loop, self-improving agent engine: Claude Code, Google Antigravity and OpenAI Codex compete for the Editor seat via blind-tournament judging, a deterministic referee (Haskell decision kernel + verified Python fallback) polices the flow, and git + an objective fitness signal arbitrate every change. Persistent sessions, context-rotation memory, AST code graph, temporal experiment knowledge graph.

Provider-neutral tool definitions and execution for Python agents

Multimodal retrieval engine: index mixed assets (PDFs / Office docs / images), search across four routes (text→text, text→image, image→image, hybrid) fused with RRF, with an optional grounded LLM answer layer on top.

A local, teaching-first RAG pipeline (Ollama + LangChain + FAISS) that turns any model into an instant expert.

Server-side SDK for the MobileOPS Managed Agent API.

Composable Python runtime for building production AI agents

MOSAIC: evaluating the Moral, Social and Individual dimensions of LLMs

A local citation-aware research and teaching assistant for technical PDFs