Tag
A new book by Gigi Sayfan guides readers on building multi-agent AI systems from scratch using Python, MCP, and A2A protocols, focusing on custom orchestration rather than third-party frameworks.
EvoScientist is an open-source framework that automates research workflows using self-evolving AI scientists with persistent multi-agent memory, adopting a human-on-the-loop paradigm for autonomous research exploration and insight generation.
Caliby is an open-sourced embedded vector database co-developed by Sea-Land AI and MIT's Michael Stonebraker team, offering high-performance vector retrieval (4x faster than pgvector) with HNSW, DiskANN, and IVF+PQ indexes, designed specifically for AI Agent and RAG use cases with a simple pip install.
CrewAI released a checkpointing feature for its open-source multi-agent framework, allowing AI workflows to be saved, resumed, forked, and inspected rather than restarting from scratch on failure.
This article explains how to build a Claude agent using Python, emphasizing the importance of handling tool failure cases effectively rather than just relying on happy-path scenarios.
Mojo, the high-performance programming language from Modular, has reached version 1.0.0 beta 1.
Release of OpenAI Python library version 2.35.1, providing updated access to the OpenAI REST API with support for workload identity authentication.
The author details the development of 'Deptool', a custom Python-based deployment and configuration management tool designed to be faster and more predictable than existing solutions like Ansible, driven by a desire for digital sovereignty and better tooling.
This is the README for anthropics/anthropic-sdk-python v0.98.1, a Python client library for accessing the Claude API.
A minimal Python ctypes binding to the TRE regex library demonstrates robust performance against ReDoS attacks, processing evil patterns on 10M-character inputs much faster than Python's re module.
The OpenAI Python client library has been updated to version 2.34.0, introducing workload identity authentication support for secure environments like Kubernetes, Azure, and GCP.
Release announcement for llm version 0.32a1, a Python CLI and library for working with Large Language Models, by Simon Willison.
LLM 0.32a0 is a major backwards-compatible refactor of the Python library and CLI tool, shifting from simple text prompts to supporting sequences of messages and multi-part responses to better handle modern LLM capabilities like structured JSON and tool use.
Developer shares experience switching Python projects from uv to PDM, highlighting PDM’s pure-Python codebase, new 2.26.8 release with relative-time dependency cooldown, and enhanced project-management features.
DSPy 3.2.0 improves dspy.RLM parsing, tool execution, and failure recovery, plus ongoing work to decouple from LiteLLM.
NeuralSet is a new Python package that provides fast, scalable preprocessing and embedding tools for multimodal neuro-AI data including fMRI, EEG, MEG, ECoG, spikes, plus text, audio, video and images.
Faceoff is a terminal UI application for following NHL hockey games in real-time, featuring live scores, game details, standings, player stats, and team information with a responsive layout.
A beginner-friendly Python package and CLI tool to control keyboard backlight brightness on Linux with mock-safe testing.
PyCon US 2026 is coming to Long Beach, California (May 13-19) with new dedicated AI and Security tracks. The AI track features talks on AI-assisted development, LLMs on laptops, voice agents, and Python async patterns for AI applications.
Datasette 1.0a28 alpha release fixes compatibility bugs and resource management issues discovered in the previous alpha, including fixes for execute_write_fn() callbacks, database cleanup methods, and a new pytest plugin for automatic cleanup in tests.