PythonHub

dottxt-ai / outlines Structured Outputs https://github.com/dottxt-ai/outlines

Adrian Open-source runtime AI agent security tool - monitors and controls AI agents, catching malicious tool use, prompt injection, and policy drift in real time, before the agent acts. https://github.com/secureagentics/Adrian

Self-contained highly-portable Python distributions https://gregoryszorc.com/docs/python-build-standalone/main/

XY – Fast, composable, GPU-accelerated Python charting library https://github.com/reflex-dev/xy

MappingTools. Do stuff with Mappings and more. This library provides utility functions for creating, manipulating, and transforming data structures, which have or include Mapping-like characteristics. https://erivlis.github.io/mappingtools/

Scaling NumPy on Free-Threaded Python A recap on the work done in NumPy and CPython to make multi-threaded NumPy workloads scale on the free-threaded build of CPython. https://labs.quansight.org/blog/scaling-numpy-on-free-threaded-python

AlgebraX - Algebraic Structures in pure Python AlgebraX is a Python library for generalized algebraic structures, focusing on Semirings and their applications in Graph Theory, Signal Processing, and Physics. https://github.com/erivlis/algebrax

Pip 26.2: –only-deps solves 16 years of app deployment hacks Pip 26.2 is expected to add pip install --only-deps ., allowing developers to install dependencies from pyproject.toml without installing the project itself. The feature resolves a 16-year pain point for deploying Python applications and scripts, replacing brittle requirements files, editable installs, and third-party workarounds. https://jamesoclaire.com/2026/07/23/pip-26-2-only-deps-solves-16-years-of-app-deployment-hacks/

CheapSecurity Provides a lightweight, self-hosted CCTV solution designed for Linux-based single-board computers (SBCs) and standard USB webcams https://github.com/gmrandazzo/CheapSecurity

Your Codebase Is Too Fragile The video refactors a travel-booking workflow to show how structural coupling, temporal coupling, and connascence can make software fragile. It explains how to identify and reduce these dependencies so future changes are safer and the code is easier to maintain https://www.youtube.com/watch?v=UEqk0njCuQo

Rune 1.1: adds Python, an Emacs editor, a symbol index and is now free https://rune.build/blog/rune-1-1-release

blader / humanizer Agent skill that removes signs of AI-generated writing from text https://github.com/blader/humanizer

Local text, image, video, music and 3D from one CLI, no Python https://github.com/sawfwair/mere-run

AgriciDaniel / claude-seo Universal SEO skill for Claude Code. 25 sub-skills + 18 sub-agents covering technical SEO, E-E-A-T, schema, GEO/AEO, backlinks, local SEO, maps intelligence, semantic clustering, e-commerce SEO, international SEO, Google APIs, and PDF/Excel reporting. Optional DataForSEO, Firecrawl, and Banana extensions. https://github.com/AgriciDaniel/claude-seo

harbor-framework / harbor Framework for evaluating and improving agents https://github.com/harbor-framework/harbor

The Design of Everyday Cryptography The article explains why cryptographic protocols should be designed to be difficult to misuse, comparing the simplicity of key encapsulation mechanisms (KEMs) with the complexity of modern digital signatures. It proposes designing opinionated, single-purpose, testable, and upgradeable cryptographic interfaces that reduce implementation errors and improve security. https://www.dlp.rip/everyday-cryptography/

A tool for finding the causes of unstable Python tests https://github.com/mgaitan/pytest-leak-finder

i-have-adhd A skill for your coding agent to stop it from burying the answer. ADHD-friendly output. https://github.com/ayghri/i-have-adhd

Bringing PyTorch Monarch to AMD GPUs: Single-Controller Distributed Training on ROCm PyTorch Monarch now supports AMD GPUs through ROCm, enabling developers to orchestrate large distributed training workloads from a single Python program. Its fault-tolerant architecture isolates failures and lets healthy GPUs continue training while failed nodes recover, reducing restarts, wasted computation, and infrastructure costs. https://pytorch.org/blog/bringing-pytorch-monarch-to-amd-gpus-single-controller-distributed-training-on-rocm/

13 Models and 4 Agents on SWE Tasks: Go, Java, Python, Rust, TS https://swe-rebench.com