Design
Preset
Background
Text
Font
Size
Width
Account Pricing Wednesday, July 22, 2026

The Git Times

“Whether a technology is liberating or enslaving depends on who controls it.” — Ursula Franklin

AI Models
Claude Fable 5 $50/M GPT-5.6 Luna $6/M Gemini 3.1 Pro Preview $12/M Grok 4.5 $6/M DeepSeek V4 Pro $0.87/M Qwen3.7 Max $4.42/M Kimi K3 $15/M
Full Markets →
Fresh on Hugging Face

Model Drops

The newest model releases builders are picking up right now.
Fresh on GitHub

Just Shipped

Significant new releases from the AI and dev-infra repos builders run on.

MovieBox-Tui Brings Streaming to the Terminal with Rust Speed 🔗

A keyboard-driven TUI fetches and plays media via mpv, requiring no VPN or configuration.

mesamirh/MovieBox-Tui · Rust · 377 stars 6d old

MovieBox-Tui is a terminal user interface that streams movies and TV shows directly from the command line. Built in Rust, it leverages the MovieBox API to search, browse, and play content using mpv as the playback engine. Users launch the app with moviebox-tui, search with /, and press Enter to stream instantly—no browser, no ads, no setup.

The interface shows full metadata including seasons, episodes, dubs, and subs, and allows copying stream URLs to the clipboard. A built-in geo-unblocking feature bypasses regional restrictions without requiring a VPN, a notable advantage for users in restricted regions. Internal network logs are accessible via Ctrl+L for debugging or inspection. The project is dual-licensed under MIT or Apache-2.0, appealing to developers concerned with licensing flexibility. Its reliance on Rust ensures memory safety and performance, while the TUI design prioritizes keyboard navigation and minimal resource usage. The project is cross-platform, working wherever Rust and mpv are available. Despite its rapid rise—377 stars in six days—the tool remains early in its lifecycle.
The catch: It depends on an external, unverified streaming API whose long-term availability and legal standing are unclear, posing a risk for sustained use.

Use Cases
  • Developers streaming media in terminal-only environments
  • Users bypassing geo-restrictions without installing VPN software
  • Rust enthusiasts building or learning from a real-world TUI application

Source: mesamirh/MovieBox-Tui — based on the project README.

More on the Front Page

OpenCodex Lets Any LLM Power Codex CLI and Claude Code 🔗

The TypeScript proxy translates Codex’s API to work with Claude, Gemini, Grok, Ollama, and more — locally.

lidge-jun/opencodex · TypeScript · 3.1k stars 1mo old

OpenCodex acts as a local middleware that intercepts Codex CLI, App, and SDK requests to the OpenAI Responses API and reroutes them to any supported LLM provider. It handles streaming, tool calls, reasoning tokens, and image input in both directions, preserving the full Codex workflow while swapping the underlying model.

Users simply run ocx start after installing via npm, then point their Codex tools to localhost:10100. The proxy also manages ChatGPT account pools for Codex authentication, auto-rotating sessions to the least-used healthy account and preserving thread affinity to prevent mid-conversation account jumps. Built in TypeScript, it supports macOS (arm64/x64) and Linux (x64/arm64) with launchd and systemd integration. The latest release, v2.7.33, is a version bump from the preview, indicating stabilization after recent feature work. Developers avoid vendor lock-in and can experiment with models like DeepSeek, Qwen, or self-hosted Ollama instances without waiting for official Codex integrations.
The catch: Windows support is incomplete in the README, and with 19 open issues and only one month of public availability, long-term reliability and enterprise-grade scalability remain unproven.

Use Cases
  • Run Codex CLI with Gemini via local proxy
  • Power Claude Code using DeepSeek reasoning
  • Test Ollama models with Codex SDK workflows

Source: lidge-jun/opencodex — based on the README and release notes.

Agent Execution Partnership adds safety gates to AI agent actions 🔗

Open-source Python control plane enforces authorization, observation, and verification for every move

eli-labz/Agent-Execution-Partnership · Python · 251 stars 1d old

Agent Execution Partnership (AEE) is a new open-source control plane that intercepts AI agent proposals, checks them against policy and risk rules, executes bounded actions via adapters, and verifies outcomes. It enforces default deny, mandatory post-action verification, and immutable audit trails.

Human approval gates trigger for high-risk operations. The v0.1.0 release includes autoresearch demos and a local site visualizing the closed loop. Built in Python, it supports pluggable adapters and MLOps integration.
The catch: Production-impact adapters are disabled by default, limiting real-world use until teams enable and test them.

Use Cases
  • Dev teams auditing LLM agent actions in staging
  • Security engineers enforcing least-privilege AI workflows
  • MLOps pipelines adding verification to autonomous task execution

Source: eli-labz/Agent-Execution-Partnership — based on the README and release notes.

open·kritt Orchestrates AI Agents to Find Real Code Vulnerabilities 🔗

Built by top bug bounty hunters, it turns focused AI scans into ranked, validated findings.

Kritt-ai/open-kritt · JavaScript · 283 stars 1d old

open·kritt breaks security research into small tasks, runs them across AI agents in parallel, and combines outputs into de-duplicated, ranked findings. Researchers build workflows, scan local or remote repos with Codex or Claude Code, verify issues via post-scripts, and prioritize results using custom severity rankers.

It requires Git, Docker, and Node.js 20+, with setup via ./kritt start. The platform is the open-source distillation of internal tools that earned over $1.5M in bug bounties under the name Blockian. The catch: As a 2-day-old project with 16 open issues, its long-term stability and scalability for enterprise use remain unproven.

Use Cases
  • Security researchers validate AI-findings in personal repos
  • Developers automate dependency scanning in CI pipelines
  • Teams build custom security playbooks for internal audits

Source: Kritt-ai/open-kritt — based on the README and release notes.

Live Tennis API MCP Server Delivers Real-Time Scores to LLM Agents inheadline: JavaScript server powers real-time tennis data for Claude, Cursor, and Zed via MCP protocol. 🔗

MCP server for the Live Tennis API — give Claude, Cursor and other LLM agents real-time tennis scores, odds and model win-probability

livetennisapi/livetennisapi-mcp · JavaScript · 260 stars 3d old

The livetennisapi-mcp project provides an MCP server that connects LLM agents to live tennis scores, odds, and win-probability models. Built in JavaScript/TypeScript, it requires no local install—agents invoke it via npx -y livetennisapi-mcp with an API key from livetennisapi.

com. Twelve tools deliver live matches, player profiles, fixtures, and premium analysis tiers. The server registers /.well-known/glama.json only when MCP_MAINTAINER_EMAIL is set, preventing accidental domain claims.
The catch: As a 3-day-old project with 6 open issues, long-term reliability and scaling under heavy LLM agent load remain unproven.

Use Cases
  • Developers add live tennis scores to Claude Code agents
  • Cursor users query real-time match odds and player stats
  • Zed editors embed ATP/WTA win-probability in prompts

Source: livetennisapi/livetennisapi-mcp — based on the README and release notes.

Tuistore solves terminal app discovery with one-key installs 🔗

It maps 800+ TUIs to correct package managers for your OS automatically

Gheat1/tuistore · Python · 242 stars 3d old

Tuistore is a Python TUI app store that lets builders browse, search, and install terminal apps with one key. It detects your OS’s package manager (brew, pacman, apt, etc.

) and shows only install commands that work. It pulls from a curated list of 818 tools, seeded from awesome-tuis and expanded via cli.masoko.net. Install via one-liner, Homebrew, PyPI, or AUR. It streams install output live and lets you star apps on GitHub from within the interface. Built on ricekit and Textual, it supports Linux, macOS, and Windows.
The catch: As a four-day-old project with zero open issues, long-term maintenance and community response to breakage remain untested.

Use Cases
  • Developers quickly install CLI tools like fzf or starship
  • Sysadmins provision machines with verified terminal app sets
  • Contributors discover and star TUIs without leaving the terminal

Source: Gheat1/tuistore — based on the README and release notes.

MEV Arbitrage Bot Automates Ethereum Pool Swaps via Smart Contract 🔗

Python script controls Solidity contract to execute arbitrage in single transactions

MIgHTy-alIeN/MEV-Arbitrage-Bot · Solidity · 739 stars 4d old

The MIgHTy-alIeN/MEV-Arbitrage-Bot combines a Solidity smart contract with Python automation to find and execute arbitrage between Uniswap pools and routers. Owners deploy the contract via Etherlab, then configure allowed routers, tokens, and swap limits through functions like setRouterAllowed() and setMinQuickSwapAmount().

The core executeArbitrage() function scans and acts on opportunities in one transaction, using the contract’s ETH/token balance. Withdrawals and approvals are owner-only, and an emergency setPaused() halts all activity. The bot has gained 739 stars and 497 forks in five days, signaling rapid adoption.
The catch: No audits or test coverage are mentioned, raising security concerns for real-value deployment.

Use Cases
  • Developers automate cross-pool arbitrage on Ethereum
  • Traders exploit price inefficiencies between Uniswap routers
  • Owners manage bot settings and withdraw funds securely

Source: MIgHTy-alIeN/MEV-Arbitrage-Bot — based on the project README.

Rust Runtime Streamlines AI Agent Repository Access 🔗

FastCtx reduces tool overhead for coding agents via structured MCP commands

yc-duan/fastctx · Rust · 308 stars 5d old

FastCtx provides a persistent Rust runtime that simplifies repository operations for AI agents. It offers structured read, grep, glob, and replace tools through the Model Context Protocol, eliminating the need for agents to construct and parse shell commands.

By handling path resolution, encoding, and output boundaries internally, it reduces context waste and lets models focus on code understanding and planning. The tool supports ChatGPT App and Codex CLI out of the box, with fastctx serve enabling any MCP client to integrate. The catch: As a v0.1.1 release just five days old, long-term reliability and real-world agent performance remain unproven at scale.

Use Cases
  • AI agents reading source files efficiently
  • Coding agents searching codebases with grep
  • Agents performing batch replacements safely

Source: yc-duan/fastctx — based on the README and release notes.

Open Source AI Agents Shift from Assistants to Autonomous Task Orchestrators 🔗

New projects reveal a pattern of agents executing complex workflows with authorization, skill registries, and multimodal output generation

The open source AI agent landscape is evolving beyond simple LLM wrappers toward systems that orchestrate autonomous, multi-step actions. Projects like eli-labz/Agent-Execution-Partnership introduce control planes that enforce pre-action authorization, real-time observability, and post-execution verification — critical for safe agent deployment in regulated environments.

Skill-based architectures are gaining traction: iflytek/skillhub offers a self-hosted registry for versioned agent skills with RBAC and audit logs, enabling enterprise-scale skill sharing, while kangarooking/cangjie-skill distills books and podcasts into executable agent skills. Multimodal agent capabilities are expanding: calesthio/OpenMontage provides a full agentic video production system with 12 pipelines and 500+ skills, turning AI assistants into studios, and gnipbao/story-to-handdrawn-video converts Chinese narratives into hand-drawn comic animations. Workflow orchestration is maturing — triggerdotdev/trigger.dev enables fully managed AI agent workflows, and stablyai/orca serves as an ADE for fleets of parallel agents across desktop and mobile. Even UIs are adapting: Jakubantalik/thinking-orbs delivers tuned loading states for agent interfaces, and alibaba/page-agent allows natural language control of web GUIs. These projects signal a shift from conversational agents to reliable, skill-driven, observable systems capable of executing complex, real-world tasks.
The catch: Despite rapid innovation, the ecosystem remains fragmented — skill formats, orchestration protocols, and security models lack standardization, making interoperability between agents uncertain and raising questions about whether today’s demos will scale to production reliability without clearer foundations.

Use Cases
  • Developers build regulated financial trading agents with audit trails
  • Enterprises deploy private agent skill libraries with access controls
  • Creators automate video production from text using agent skill chains

Open Source LLM Tools Evolve Beyond Prompt Chaining Into Agent Orchestration 🔗

Projects now standardize how LLMs interact with data, APIs, and execution environments through unified control planes and skill frameworks.

The emerging pattern in open source LLM tooling shifts from isolated prompt engineering to structured agent orchestration. Repos like eli-labz/Agent-Execution-Partnership introduce control planes that authorize, observe, and verify AI agent actions — treating LLMs as programmable components within auditable workflows.

Similarly, lidge-jun/opencodex acts as a universal provider proxy, enabling Claude Code and OpenAI Codex to route requests through any LLM backend, abstracting model choice from tooling. Skill development is also consolidating: microsoft/SkillOpt trains reusable natural-language skills for frozen LLMs via trajectory-driven edits, while AgriciDaniel/claude-seo packages 25 SEO sub-skills and 18 sub-agents into a modular framework for Claude Code. Data integration follows suit — livetennisapi/livetennisapi-mcp exposes real-time tennis data via an MCP server, letting LLM agents access live scores and odds as first-class inputs. Execution environments are hardening too: deer-flow/llm-space offers a desktop app to prototype, replay, and evaluate agent workflows locally, and mukul975/Anthropic-Cybersecurity-Skills maps 754 structured skills to frameworks like MITRE ATT&CK for security-focused agents.
The catch: Much of this infrastructure remains fragmented across incompatible agent protocols (MCP, A2A, custom harnesses), with limited interoperability between skill libraries and control planes — promising in isolation but unproven at scale in production agent swarms.

Use Cases
  • Developers build auditable trading agents with authorization logs
  • Security teams deploy LLM agents using standardized cybersecurity skill sets
  • Data analysts connect LLMs to real-time sports APIs for dynamic insights

Open Source Frameworks Converge on AI-Native Workflows 🔗

New projects blend LLMs, real-time data, and modular design to reshape how developers build intelligent applications

A clear pattern is emerging in open source web frameworks: the shift from static CRUD tools to AI-integrated, event-driven platforms that treat LLMs as first-class citizens. Projects like twentyhq/twenty and oomol-lab/open-connector expose SaaS APIs to AI agents via MCP and SDKs, enabling autonomous workflows.

triggerdotdev/trigger.dev goes further, offering fully managed AI agent deployment with built-in scheduling and error handling. Meanwhile, alibaba/page-agent and Skyvern-AI/rustwright demonstrate in-page and browser automation driven by natural language, letting LLMs interact directly with UIs. Real-time data pipelines are also being weaponized for AI—see livetennisapi/livetennisapi-mcp feeding live sports odds to models, and JesusRS1/stock-trade-finance-api streaming financial data into LLM contexts. Even infrastructure is adapting: juspay/hyperswitch adds intelligent routing for payment flows, while mock-server/mockserver-monorepo now mocks AI/LLM APIs for testing. These aren’t just wrappers; they represent a new architectural layer where orchestration, data flow, and AI reasoning are co-designed. The catch: Much of this remains experimental—agent reliability is unproven at scale, MCP adoption is fragmented, and many projects trade maturity for novelty, leaving integrators to wrestle with unstable APIs and unclear operational boundaries.

Use Cases
  • Developers automate CRM updates using AI agents
  • QA teams test LLM-powered apps with mock APIs
  • Finance teams embed real-time market data into LLMs

Deep Cuts

Russian AI Agent Gets Authentic Slang Mode 🔗

Pohuy injects idiomatic Russian profanity for more natural, effective agent communication

smixs/pohuy · Unknown · 493 stars

smixs/pohuy is a niche but fascinating tool that equips AI agents with a mode for idiomatic Russian profanity—think of it as a linguistic seasoning for more authentic, emotionally resonant interactions in Russian-language contexts. Designed for developers building agents that need to sound genuinely local, it goes beyond textbook language to capture colloquial intensity and cultural nuance.

While labeled 18+, its value lies in pragmatic effectiveness: agents using appropriate register can build rapport faster in informal or high-stakes scenarios. The project’s quiet growth—133 new stars recently—suggests a growing appetite for culturally grounded AI behavior, even if the approach remains unconventional.
The catch: It's highly specialized, early-stage, and assumes familiarity with Russian linguistic subtleties that may limit broader adoption.

Use Cases
  • Russian customer service bots handling informal complaints
  • Game NPCs delivering culturally authentic dialogue
  • Social simulation agents modeling real-world vernacular interactions

Source: smixs/pohuy — based on the project README.

AI Love Coach Blends Psychology, Law, and Ethics 🔗

Open-source agent offering actionable relationship advice rooted in multidisciplinary knowledge

powerycy/goutoujunshi · Python · 416 stars

powerycy/goutoujunshi is a Python-based Codex skill that acts as an AI relationship advisor, first validating emotions, then analyzing dynamics, and finally delivering practical strategies. It draws from a curated knowledge base spanning psychology, law, sociology, philosophy, family studies, and sexology to support diverse relationship structures, including LGBTQ+ dynamics.

Unlike generic chatbots, it prioritizes emotional grounding before advice, making interactions feel more human and context-aware. Built as an extensible agent, it invites developers to plug into its reasoning pipeline or adapt its frameworks for ethical AI guidance in sensitive domains.
The catch: It’s early-stage, primarily Chinese-language, and lacks broad testing outside niche cultural contexts.

Use Cases
  • Developers building culturally aware AI counselors
  • Researchers testing multidisciplinary agent architectures
  • Teams creating inclusive relationship support tools

Source: powerycy/goutoujunshi — based on the project README.

Quick Hits

twenty TwentyHQ/Twenty: An open-source CRM alternative to Salesforce, built for AI integration with TypeScript, enabling customizable, intelligent customer relationship management without vendor lock-in. 53.4k
hyperswitch Juspay/Hyperswitch: A Rust-based, PCI-compliant payments platform offering composable, multi-provider connectivity for payments, payouts, fraud, and vault services — with intelligent routing, revenue recovery, cost observability, and automated reconciliation to cut ops and expenses. 43.4k
thinking-orbs Jakubantalik/Thinking-Orbs: A TypeScript library providing six tuned states and two sizes of dotted thought-orb loading indicators for AI and agent UIs, with automatic dark/light mode support for seamless, intuitive user feedback. 325
agent-notch RealFishSam/Agent-Notch: A Swift-based open-source alternative to Vibe-Island, designed to deliver immersive, interactive agent UI experiences with native performance and clean, customizable visual feedback. 235
story-to-handdrawn-video Gnipbao/Story-to-Handdrawn-Video: A JavaScript agent skill that transforms Chinese story text or ordered images into silent, hand-drawn diary-comic MP4 animations — turning narratives into expressive, sketch-style visual stories automatically. 282
Beyond GitHub

The AI Wire

What builders are reading today — the headlines, papers, and announcements that aren't trending repos.

From the labs & arXiv

OpenClaw v2026.7.1 Overhauls Onboarding and Control UI for Personal AI Assistants 🔗

The release adds GPT-5.6 support, improves cross-platform apps, and refines workspace tools for local-first AI workflows.

openclaw/openclaw · TypeScript · 383.8k stars 7mo old · Latest: v2026.7.1

OpenClaw’s latest release, v2026.7.1, delivers a complete overhaul of its onboarding flow and Control UI, making setup faster and daily use more intuitive.

The guided openclaw onboard CLI now validates connections in real time and preserves progress if interrupted, reducing friction for new users. Once running, the redesigned Control UI lets builders view conversations side by side with live Tasks, file previews, and cost metrics—all without leaving the chat interface.

Official iOS, Android, and macOS apps received substantial updates, including improved voice input, offline message queuing, and native session controls. Model support expanded to include GPT-5.6, Tencent Hy3, and Meta Muse Spark 1.1, strengthening its appeal for developers who want to plug in trusted, cutting-edge LLMs while keeping data local. Gateway daemon stability was also addressed, with fixes for crash loops and improved handling of scheduled work and remote browser control.

The project continues to emphasize a single-user, self-hosted assistant that runs on personal devices via Node.js (v24.15+ recommended), bridging WhatsApp, Slack, Telegram, and 20+ other channels through a unified workspace. It remains MIT-licensed and driven by the OpenClaw Foundation, with over 3,000 contributions in this release alone from 532 developers.

The catch: Despite its broad channel support, OpenClaw still lacks native voice output on Linux desktop and relies heavily on a stable Node.js runtime, which may complicate deployment in minimal or container-optimized environments where developers prefer lighter-weight alternatives like Python or Rust-based agents.

Previously in The Times “covered” — Jul 19

Use Cases
  • Developers syncing AI chats across WhatsApp, Slack, and Telegram
  • macOS/iOS users running a local AI assistant with voice and file handling
  • Linux users self-hosting a personal coding agent via terminal and Discord

Source: openclaw/openclaw — based on the README and release notes.

More Stories

Firecrawl v2.11.0 adds keyless access for core endpoints 🔗

New release removes API key requirement for /scrape, /search, /interact, and /parse in official clients

firecrawl/firecrawl · TypeScript · 154.2k stars Est. 2024

Firecrawl’s latest release enables keyless access to core endpoints when using its official MCP, CLI, or SDK clients, lowering the barrier for quick prototyping and agent integration. The update also introduces a Research Index for arXiv and GitHub code, automatic PII redaction, and deterministic JSON output to reduce LLM token use and cost.

Video discovery now works on any page, not just embedded platforms.
The catch: Keyless access is limited to official clients, meaning custom or third-party integrations still require API key management and rate limit handling.

Previously in The Times “covered” — Jul 16

Use Cases
  • AI agents scrape live web data without key setup
  • Developers test scraping logic in local environments fast
  • Research bots extract arXiv papers via dedicated index

Source: firecrawl/firecrawl — based on the README and release notes.

Microsoft's AI curriculum updates Jupyter labs for TensorFlow and PyTorch 🔗

Recent commit refreshes 24-week deep learning course with current framework syntax

microsoft/AI-For-Beginners · Jupyter Notebook · 52.6k stars Est. 2021

The microsoft/AI-For-Beginners project received a commit one day ago, updating Jupyter Notebook lessons for TensorFlow 2.x and PyTorch 2.

0 compatibility. This 12-week, 24-lesson curriculum covers CNNs, RNNs, GANs, and NLP with hands-on labs and quizzes. Designed for beginners, it uses Python to teach core AI concepts through practical exercises. The repository maintains 52,580 stars and sees active community forks, indicating sustained use in self-paced learning environments.
The catch: Open issues number six, with no recent response from maintainers on reported lab environment setup problems affecting Windows users.

Use Cases
  • Self-taught developers learning foundational deep learning
  • Educators adapting beginner AI coursework for classrooms
  • Students practicing TensorFlow and PyTorch via guided notebooks

Source: microsoft/AI-For-Beginners — based on the project README.

Gemini Cookbook Adds Nano-Banana 2 for Image Generation 🔗

New model enables 512px image creation with search grounding and thinking modes

google-gemini/cookbook · Jupyter Notebook · 17.6k stars Est. 2024

The Google Gemini cookbook updated its examples to include Nano-Banana 2, a 512px image generation model supporting search grounding and thinking mode. Developers can now follow step-by-step guides to generate and edit images using natural language prompts within the Gemini API framework.

The cookbook also covers Gemini 3.6 Flash for agentic tasks and Lyria 3 for music generation, reflecting recent model releases. All examples run in Jupyter Notebooks, requiring an API key and basic Python familiarity.
The catch: The cookbook assumes prior Gemini API knowledge and offers limited troubleshooting for failed generations or quota errors, leaving newcomers to rely on external documentation for debugging.

Use Cases
  • Developers generate product images from text prompts
  • Researchers test multimodal agent workflows
  • Creators prototype music-to-image pipelines

Source: google-gemini/cookbook — based on the project README.

Quick Hits

ultralytics Ultralytics YOLO models deliver real-time object detection, segmentation, pose estimation, and tracking in a unified framework for vision AI applications. 59.7k
supabase Supabase provides a fully managed Postgres database with instant APIs, auth, and storage to accelerate full-stack and AI app development. 106.8k
generative-ai-for-beginners This hands-on Jupyter notebook guide teaches beginners to build generative AI applications through 21 practical lessons covering LLMs, prompts, and chains. 113.4k
OpenBB OpenBB offers an open-source data platform integrating financial datasets, AI agents, and analytics tools for quants and developers to build custom workflows. 70.9k
langchain LangChain enables developers to compose, debug, and deploy LLM-powered agents and chains with modular components for reasoning, tool use, and memory. 142.3k

Newton Physics Engine Gains Deterministic GPU Workflows for Robotics Research 🔗

v1.4.0 enables bit-exact rollouts and solver composition using NVIDIA Warp integration for reproducible simulation

newton-physics/newton · Python · 5.2k stars Est. 2025 · Latest: v1.4.0

Newton v1.4.0 introduces experimental deterministic execution paths for XPBD, SemiImplicit, Featherstone, VBD, and MuJoCo solvers, enabling bit-exact repeated rollouts critical for reinforcement learning and control validation.

The release expands solver composition with SolverCoupled, allowing researchers to combine MuJoCo, Kamino, VBD, XPBD, and MPM within a single workflow using proxy coupling and ADMM contacts. Deformable USD import now supports curves, tetrahedral meshes, and attachments as cables, cloth, or soft bodies via ModelBuilder.add_usd(), aligned with AOUSD schema proposals. Graph-captured workloads see performance gains, and Warp graphs can now be replayed on CPU for portability. Examples like basic_pendulum and basic_urdf demonstrate rapid prototyping with pip install "newton[examples]".

The catch: Deterministic modes remain experimental and opt-in, with no guarantee of bit-exactness across all solver combinations or hardware configurations, limiting immediate drop-in use for production pipelines requiring strict reproducibility.

Previously in The Times “covered” — Jul 20

Use Cases
  • Roboticists validating control policies with repeatable simulation
  • Researchers composing multi-physics deformable and rigid dynamics
  • Teams exporting OpenUSD scenes for NVIDIA Omniverse integration

Source: newton-physics/newton — based on the README and release notes.

More Stories

OpenBot Playground adds Flutter controller with WebRTC video feed 🔗

New release enables cross-platform teleoperation via remote web server and Wi-Fi AP sharing

ob-f/OpenBot · Swift · 3.4k stars Est. 2020

The OpenBot project released v0.8.

0, introducing a Flutter-based controller for Android and iOS that streams smartphone camera feeds using WebRTC. Users can now teleoperate robots through a remote web server or share Wi-Fi hotspots between robot and controller phones. Blockly updates improve iOS and Android compatibility in the visual programming Playground. Despite six years of development, the project maintains slow-burn traction with 39 open issues and a last commit just days ago. The catch: reliance on smartphone hardware limits sensor fidelity and real-time performance compared to dedicated robotics platforms.

Use Cases
  • Educators build low-cost robots for classroom AI experiments
  • Hobbyists prototype person-following robots using Android phones
  • Researchers test autonomous navigation algorithms on $50 robot bodies

Source: ob-f/OpenBot — based on the README and release notes.

LiDAR-Camera Calibration Toolbox Gains ROS2 Support and Docker Ease 🔗

Target-less, single-shot method now runs on ROS2 with prebuilt containers for faster setup

koide3/direct_visual_lidar_calibration · C++ · 1.5k stars Est. 2022

The direct_visual_lidar_calibration project updated its Docker image and ROS2 compatibility last month, streamlining deployment for roboticists. It calibrates LiDAR and camera extrinsics without targets using scene texture and structure, requiring only one LiDAR point cloud and image pair.

Built in C++, it leverages GTSAM, Ceres, and SuperGlue for pixel-level registration, claiming higher accuracy than edge-based methods. The MIT-licensed toolbox supports spinning and non-repetitive LiDARs with pinhole, fisheye, or omnidirectional cameras.
The catch: With 72 open issues and no release in over a year, long-term maintenance and ROS2 Foxy/Humble stability remain uncertain for production systems.

Use Cases
  • Autonomous vehicle teams calibrating sensors without calibration targets
  • Robotics researchers validating extrinsic parameters in unstructured environments
  • ROS2 developers integrating LiDAR-camera sync via prebuilt Docker containers

Source: koide3/direct_visual_lidar_calibration — based on the project README.

Makelangelo software adds real-time editor updates for plotter art 🔗

New release fixes turtle rendering and travel move color issues in Java-based tool

MarginallyClever/Makelangelo-software · Java · 425 stars Est. 2012

MarginallyClever/Makelangelo-software released version 7.78.

6 with improved editor responsiveness. The update adds an EditorContext to track drawing changes and enables real-time previews during rotate, scale, and translate actions. A bug fix corrects turtle render positioning relative to the drawing origin, and the first travel move now displays in the correct color instead of white. The Java application prepares vector art for CNC plotters, particularly the wall-hanging Makelangelo polargraph, and integrates with Marlin firmware. Builders use it to generate toolpaths from SVG or raster images for precise pen plotting.
The catch: Development shows signs of stagnation with 24 open issues and no major feature additions in recent releases.

Use Cases
  • Artists convert sketches to plotter-ready G-code
  • Educators teach coordinate geometry via pen drawing
  • Makers prototype vector art on vertical surfaces

Source: MarginallyClever/Makelangelo-software — based on the README and release notes.

Quick Hits

habitat-sim A high-performance C++ 3D simulator enabling realistic embodied AI research with flexible scene and agent control. 3.8k
robot_descriptions.py A Python tool providing instant access to 190+ robot models across major robotics frameworks for rapid prototyping and simulation and control. 796
autoware An open-source autonomous driving stack in Docker, offering perception planning and control for real-world self-driving vehicle development. 11.9k
ros2_documentation The official ROS 2 documentation repository, delivering clear guides and references for building robust distributed robotics systems. 963
drake A C++ framework for model-based robotics design enabling dynamics simulation control and verification with rigorous mathematical foundations. 4.1k

Wazuh Hardens Cluster Security Fixes Address Critical Path Traversal Risks 🔗

Latest release patches file synchronization flaws that could allow directory escape in distributed deployments

wazuh/wazuh · C++ · 16.2k stars Est. 2015 · Latest: v4.14.6

Wazuh’s v4.14.6 release focuses on securing its manager cluster communications, addressing multiple path validation gaps that could enable attackers to traverse directories during file synchronization.

Fixes include improved validation for merged and non-merged file paths, temporary file handling in the cluster DAPI, and node name format checks in the hello handler—all aimed at preventing directory escape exploits. The update also resolves a segfault in the vulnerability scanner during shutdown when disabled and enhances message decompression handling in the remoted service.

These changes strengthen Wazuh’s core architecture for distributed environments, where agents report to a clustered management server processing logs, file integrity data, and vulnerability scans. The platform continues to integrate with the Elastic Stack for alert visualization and analysis, using rule-based engines to detect intrusions, misconfigurations, and policy violations across cloud, container, and on-prem workloads.

The catch: Despite its broad feature set, Wazuh’s C++ agent can be resource-intensive on constrained endpoints, and its rule tuning requires significant expertise to avoid alert fatigue in high-volume environments.

Previously in The Times “covered” — Jul 21

Use Cases
  • Securing Kubernetes clusters with file integrity and log monitoring
  • Detecting ransomware via behavioral anomaly and syscall analysis
  • Meeting PCI-DSS audit requirements through centralized log retention and correlation

Source: wazuh/wazuh — based on the README and release notes.

More Stories

OWASP Juice Shop Adds Node.js v26 Support for Modern Build Pipelines 🔗

Latest release ensures compatibility with current JavaScript runtime versions across platforms

juice-shop/juice-shop · TypeScript · 13.5k stars Est. 2014

The OWASP Juice Shop project updated its release pipeline to include Node.js v26, enabling packaged artifacts for Linux, Windows, and macOS.

This change supports developers using modern JavaScript toolchains while maintaining the application’s role as a deliberately insecure web app for security training. The project continues to encompass vulnerabilities from the OWASP Top Ten and real-world flaws, used in CTFs, demos, and tool testing. Installation remains straightforward via source, Docker, or packaged distributions.
The catch: Despite active maintenance, the project’s reliance on specific AI/LLM providers for certain challenges may limit offline or air-gapped deployment scenarios.

Use Cases
  • Security teams training developers on OWASP Top Ten vulnerabilities
  • Educators conducting hands-on application security workshops
  • Researchers testing dynamic application security testing (DAST) tools

Source: juice-shop/juice-shop — based on the README and release notes.

Strix AI pentester gains sandbox fixes in v1.2.0 release 🔗

Latest update resolves transient startup failures and caps container logs to prevent disk exhaustion

usestrix/strix · Python · 43.3k stars 11mo old

The Strix project released v1.2.

0, addressing runtime instability in its Docker-based AI penetration testing agents. Fixes include retry logic for sandbox startups, shared network port resolution, and memory bounds per agent. A reverted change and new log-cap feature now prevent host disk exhaustion during scans. Developers run Strix via CLI after setting `strix --target ./app after installing with Docker and an LLM API key. It validates exploits with real PoCs, not just static flags, aiming to reduce false positives in CI/CD pipelines. The catch: Open issues remain high at 215, raising questions about long-term reliability despite frequent commits.

Previously in The Times “covered” — Jul 18

Use Cases
  • Security teams automate vuln scans in PRs
  • Developers validate exploits with real PoCs
  • Bug hunters accelerate bounty reporting workflows

Source: usestrix/strix — based on the README and release notes.

KeePassXC Adds Passkey Support With Potential Breaking Changes 🔗

Latest release updates FIDO2 flags and fixes Auto-Type race conditions on Linux

keepassxreboot/keepassxc · C++ · 28.1k stars Est. 2016

KeePassXC 2.7.

12 enables BE and BS flags for passkeys, which may break existing credentials. The release adds TIMEOTP autotype support, improves browser URL display, and fixes OpenSSL exploit vectors. Auto-Type now avoids a Linux-specific race condition after a recent revert. Bitwarden imports gain nested folder handling. Despite steady traction, the project’s 10.4-year age and 897 open issues raise questions about long-term maintainability.
The catch: Nearly 900 unresolved issues suggest ongoing stability risks for enterprise users relying on zero-failure password management.

Use Cases
  • Developers securing API keys across Windows, macOS, and Linux
  • Teams sharing encrypted credential stores via cloud-synced KDBX files
  • Individuals generating and auto-typing strong passwords in browsers and apps

Source: keepassxreboot/keepassxc — based on the README and release notes.

Quick Hits

osmedeus j3ssie/osmedeus (Go): A modern orchestration engine that automates security testing workflows, enabling scalable, repeatable vulnerability discovery across complex infrastructures. 6.5k
berty berty/berty (TypeScript): A secure peer-to-peer messaging app that functions offline or on unreliable networks without trusting intermediaries, ensuring private communication by design. 9.2k
hacktricks HackTricks-wiki/hacktricks (CSS): A curated, continuously updated knowledge base of practical hacking techniques from CTFs, real-world exploits, and security research for offensive security practitioners. 11.9k
authelia authelia/authelia (Go): An OpenID Certified™ SSO and MFA portal that adds strong, centralized authentication to web apps with minimal configuration and maximum security. 28.3k
sniffnet GyulyVGC/sniffnet (Rust): A user-friendly network traffic monitor that visualizes and analyzes internet activity in real time, helping users understand and control their data flow. 40.1k

React Native v0.86.0 Boosts DevTools and Animation Backend Stability 🔗

New Dev Server hooks and AccessibilityInfo fixes improve debugging and cross-platform reliability

react/react-native · C++ · 126.2k stars Est. 2015 · Latest: v0.86.0

React Native’s latest release, v0.86.0, sharpens developer experience with targeted upgrades to DevTools and the Animation Backend.

Dev Server now calls global.__BUNDLE_LOADER_REPORTER__ during Metro bundler loads, enabling custom telemetry for build performance. React Native DevTools gains a new custom track for Renderer operations, giving engineers granular visibility into UI thread activity in performance traces.

On the animation front, two commits from Bartlomiej Bloniarz introduce weak_ptr usage in pullAnimationMutations callbacks to prevent retain cycles and add Suspense-compatible tests for the Animation Backend. Accessibility sees Promise-based method tests added to AccessibilityInfo, with null checks and error messages aligned for consistency.

These changes address long-standing friction points: unstable animations under concurrent mode and opaque Dev Server behavior. Teams using custom Metro plugins or debugging jank in complex UIs will see immediate value. The release supports iOS 15.1+ and Android 7.0+, requiring macOS for iOS builds unless using Expo.

The catch: Despite these refinements, React Native still relies on a C++ bridge that can complicate native module debugging, and its large binary size remains a concern for lightweight apps where Flutter or Kotlin Multiplatform may offer leaner alternatives.

Previously in The Times “covered” — Jul 18

Use Cases
  • Mobile teams debugging UI jank in React Native apps
  • Teams adding custom Metro build telemetry
  • Developers integrating Suspense with animated components

Source: react/react-native — based on the README and release notes.

More Stories

PocketBase 0.39.8 fixes JSVM state leaks and adds Shift+Click bulk UI selection 🔗

SQLite-powered Go backend updates dependencies and patches minor security issues

pocketbase/pocketbase · Go · 60.1k stars Est. 2022

PocketBase released v0.39.

8, resetting the JSVM global $app overwrite to prevent state leaks in pooled executors. The update adds Shift+Click range bulk selection in the admin UI and normalizes number inputs with leading zeros. Dependencies like golang.org/x/* and modernc.org/sqlite (v1.54.0, SQLite 3.53.3) were bumped for security fixes. Built on a single file, it combines embedded SQLite with realtime subscriptions, file/user management, and a REST-ish API. Developers can run it as a standalone executable or embed it as a Go library.
The catch: Backward compatibility is not guaranteed before v1.0.0, posing risks for production apps relying on breaking changes.

Use Cases
  • Rapid prototyping of realtime apps with auth and file storage
  • Embedding a lightweight backend in Go-based desktop or CLI tools
  • Creating internal dashboards with instant data sync and user management

Source: pocketbase/pocketbase — based on the README and release notes.

Electron v43.2.0 Adds WHATWG WebSocket for Main Process 🔗

Fixes Windows GPU disable, Wayland shortcuts, and Linux UI crashes

electron/electron · C++ · 122.1k stars Est. 2013

Electron v43.2.

0 introduces net.WebSocket, a WHATWG-compatible client for the main process that routes through Chromium’s network stack. It also resolves a Windows GPU acceleration bug where app.disableHardwareAcceleration() failed since v38, fixes GNOME Wayland globalShortcut failures by enabling Chromium’s portal trigger, and stops Linux message box crashes when another UI layer is active. macOS window throttling under transparent covers is now fixed. Built on Node.js and Chromium, Electron powers cross-platform apps like VS Code.
The catch: 875 open issues suggest ongoing stability work, and the framework’s size and Chromium dependency remain concerns for lightweight apps.

Use Cases
  • Developers build Slack-like desktop apps with web tech
  • Teams package internal tools for Windows, macOS, Linux
  • Creators prototype APIs using Electron Fiddle for rapid testing

Source: electron/electron — based on the README and release notes.

Rust-based UV speeds Python dependency resolution with workspace path support 🔗

Latest release enables cross-workspace references and centralized virtual environment files

astral-sh/uv · Rust · 87.7k stars Est. 2023

Astral's UV 0.11.

31 adds the ability for workspace sources to reference members in another workspace by path, streamlining monorepo dependency management. It also supports .venv files pointing to centralized project environments, reducing redundant virtual environment setup time. Performance improvements avoid quadratic work when deduplicating transitive conflicts, maintaining its 10-100x speed edge over pip. The release updates Windows timezone data and introduces malware-check audit settings for enhanced security.
The catch: Despite rapid growth, UV’s extensive feature set may overwhelm teams seeking simpler, more focused tooling.

Previously in The Times “covered” — Jul 16

Use Cases
  • Dev teams managing monorepos with shared Python dependencies
  • CI/CD pipelines requiring fast, reliable environment provisioning
  • Developers replacing multiple Python tools with a single binary

Source: astral-sh/uv — based on the README and release notes.

Quick Hits

ladybird LadybirdBrowser/ladybird: A fully independent C++ web browser offering a standards-compliant, privacy-focused alternative to Chromium and Firefox engines. 64.6k
kubernetes kubernetes/kubernetes: Automates deployment, scaling, and management of containerized applications across clusters with declarative configuration and self-healing capabilities. 123.9k
tauri tauri-apps/tauri: Enables building lightweight, secure desktop and mobile apps using web frontends with Rust backends, minimizing bundle size and attack surface. 109.3k
cc-switch farion1231/cc-switch: A cross-platform Rust-based desktop assistant unifying access to Claude Code, Codex, OpenCode, OpenClaw, Grok Build, and Hermes Agent via a single interface. 120k
deno denoland/deno: A secure, modern runtime for JavaScript and TypeScript with built-in tooling, ES modules, and permission-based execution — no Node.js compatibility required. 107.8k

TuyaOpen v1.9.0 Adds Ultra-Low Power Wi-Fi Mode for Battery Devices 🔗

New DTIM support and board expansions target longer runtime in AIoT hardware designs

tuya/TuyaOpen · C · 1.8k stars Est. 2023 · Latest: v1.9.0

TuyaOpen’s latest release introduces Wi-Fi ultra-low-power DTIM mode, significantly cutting standby consumption for battery-operated devices via the new ultra_lowpower_demo example. The update also adds support for three T5AI platform boards—ZECTRIX_T5AI_NOTE_4, SPARKLEIOT_T5_DEV, and LCKFB—complete with integrated drivers for GC0308 camera and SC7A20 IMU sensors. Linux users gain RK3576 board compatibility, enabling direct deployment of chatbot applications like your_chat_bot.

These expansions broaden TuyaOpen’s hardware reach across Tuya T-Series, ESP32, and Linux SBCs, reinforcing its role as a C-based SDK for embedding LLMs (ChatGPT, Gemini, Deepseek) and multimodal AI into smart devices. Developers can now build voice-controlled products with ASR, KWS, TTS, and STT while maintaining Tuya Cloud connectivity for OTA updates and remote management. Despite active commits, the project shows slow traction with 73 open issues and limited recent community engagement beyond core maintainers.
The catch: Heavy reliance on Tuya Cloud’s proprietary AI workflows may lock developers into a vendor-specific ecosystem, complicating migration to open or multi-cloud alternatives.

Previously in The Times “covered” — Jul 17

Use Cases
  • Build battery-powered voice assistants with extended standby life
  • Create vision-enabled AI cameras using integrated GC0308 drivers
  • Deploy Linux-based chatbots on RK3576 edge hardware for local LLMs

Source: tuya/TuyaOpen — based on the README and release notes.

More Stories

C++ Roadmap Updates With Fresh Draw.io Editable Format 🔗

Project adds SVG and Miro exports to aid self-paced learning for developers

salmer/CppDeveloperRoadmap · HTML · 3.5k stars Est. 2021

The salmer/CppDeveloperRoadmap project recently updated its core `roadmap.drawio.

svg` file, combining editable draw.io source with SVG output in a single artifact. Hosted on GitHub, the HTML-based roadmap guides learners through C++ fundamentals to advanced topics like tooling and pet projects, using structured sections on application areas, myths, and study methods. It supports self-study via linked books, resources, and grading frameworks for junior to middle developers. The project sees slow traction with 3,509 stars and 326 forks, last updated over a year ago.
The catch: Open issues linger for four months with no recent commits beyond minor edits, raising questions about active maintenance and content freshness.

Use Cases
  • Junior devs learning C++ core syntax and best practices
  • Mid-level engineers reviewing modern C++ tooling and libraries
  • Self-study planners building structured study schedules from scratch

Source: salmer/CppDeveloperRoadmap — based on the project README.

ESP32 Pinout Studio Prevents Costly Maker Mistakes 🔗

Interactive tool maps pins with live conflict detection and exportable Arduino defines

FelixKunzJr/ESPPinoutWebsite · TypeScript · 36 stars 1mo old

FelixKunzJr/ESPPinoutWebsite offers a free, interactive pinout reference for the full ESP32 family at esp32pin.com.

Builders select modules, view schematic or physical layouts, and use a pin mapping builder that flags strapping, ADC2/Wi-Fi, flash-reserved, and input-only pin conflicts in real time. The tool exports Arduino-compatible #define statements, shareable URLs, and PNG diagrams, sourcing pin data directly from Espressif’s official KiCad libraries to ensure accuracy. Despite its utility, the project is two months old with only 36 stars and two open issues, raising questions about long-term maintenance and community adoption beyond early traction.
The catch: Early-stage maturity means limited real-world validation and uncertain support for future ESP32 variants or breaking toolchain changes.

Use Cases
  • Embedded developer validating GPIO safety before PCB design
  • Maker avoiding Wi-Fi crashes by filtering ADC2-unsafe pins
  • Educator teaching ESP32 pin constraints with live diagrams

Source: FelixKunzJr/ESPPinoutWebsite — based on the project README.

hwinfo library adds cross-platform hardware data access for C++ developers 🔗

Latest update improves GPU and memory reporting on Linux and macOS

lfreist/hwinfo · C++ · 723 stars Est. 2022

The lfreist/hwinfo project released updates enhancing hardware detection across Windows, Linux, and macOS. Developers can now query CPU vendor, model, core counts, and cache size reliably on all platforms.

GPU vendor and model reporting improved on Linux, while memory size and free size are now consistent across Apple and Windows. Disk serial numbers, volumes, and OS architecture details remain fully supported. Integration via CMake or Git submodule is documented. The catch: GPU memory size and battery vendor details are still missing on Linux and macOS, limiting full hardware visibility on those platforms.

Use Cases
  • System admins checking CPU and RAM usage on mixed OS fleets
  • Game developers tuning performance based on GPU and core counts
  • Hardware inventory tools gathering disk and motherboard details remotely

Source: lfreist/hwinfo — based on the project README.

Quick Hits

pxt-maker Enables visual programming for Maker boards via MakeCode, letting builders prototype hardware projects with drag-and-drop blocks and JavaScript/TypeScript. 129
childrens-clock A child-friendly IoT clock that teaches time concepts through interactive, colorful displays and safe, low-power wireless connectivity. 53
ghdl A robust, standards-compliant VHDL simulator supporting 2008/93/87 revisions, enabling accurate FPGA/ASIC design verification and synthesis workflows. 2.8k
PipelineC Introduces automatic pipelining as a first-class language construct in Python, letting designers express high-performance hardware with HLS-like efficiency. 737
hwloc Provides portable hardware topology discovery, helping builders optimize software performance by revealing CPU, memory, and accelerator layouts across systems. 723

Flame Engine 1.38.0 Fixes Gesture Detector Instability in Flutter Games 🔗

Breaking refactor improves input handling reliability while patching collision and disposal bugs

flame-engine/flame · Dart · 10.7k stars Est. 2017 · Latest: v1.38.0

The Flame Engine’s v1.38.0 release targets core stability with a refactor of GestureDetectorBuilder and callback registration ([#3900](https://github.

com/flame-engine/flame/issues/3900)), making input handling more resilient in Flutter-based games. This breaking change addresses flaky gesture detection that could disrupt gameplay, alongside fixes for TextBoxComponent endless loops (#3939), double-dispose risks (#3909), and inaccurate raycasting collisions (#3906). The engine retains its Flutter-first approach, offering a game loop, component system (FCS), particles, and bridge packages like flame_audio and flame_bloc for integrating audio and state management. Documentation notes that main branch content exceeds released docs, urging developers to check version selectors. Despite 86 open issues and a 0-day-old last commit signaling active maintenance, the project’s 8.8-year age raises questions about long-term architectural agility.
The catch: Flame’s tight Flutter coupling limits reuse outside the Flutter ecosystem, posing a barrier for teams targeting multi-platform game engines with shared logic.

Previously in The Times “covered” — Jul 20

Use Cases
  • Indie devs building 2D Flutter games with physics
  • Startups prototyping interactive Flutter experiences
  • Educators teaching game dev with Dart and Flame

Source: flame-engine/flame — based on the README and release notes.

More Stories

Babylon.js patches depth rendering in Gaussian Splatting workflows 🔗

Latest update fixes antialiased edge preservation and improves FBX handling in Inspector v2

BabylonJS/Babylon.js · TypeScript · 25.8k stars Est. 2013

Babylon.js 9.

17.0 refactors Gaussian Splatting mesh camera limits and retries deferred LOD switches after cooldown. A core bug fix by VicenteCartas preserves antialiased edges when writing depth, addressing rendering artifacts in splat-based scenes. The Inspector now accepts FBX files during asset authoring, streamlining 3D pipeline integration. Developers gain tighter control over splat rendering and smoother asset workflows. The catch: Open issues remain at 19, with no indication of major architectural updates despite steady traction.

Use Cases
  • Game devs render point-cloud scenes with Gaussian Splatting
  • 3D artists import FBX models directly into Babylon.js Inspector
  • WebXR builders stabilize depth rendering in immersive experiences

Source: BabylonJS/Babylon.js — based on the README and release notes.

libGDX 1.14.2 patches ClickListener and font cache bugs 🔗

Minor release improves UI reliability and text rendering across platforms

libgdx/libgdx · Java · 25.2k stars Est. 2012

libGDX 1.14.

2 fixes a stuck button state when ClickListener is cancelled and resets glyphCount in BitmapFontCache.clear(). It also reverts addAll to return void and adopts exact Delaunay triangulation for better mesh generation. These changes address edge cases in 2D UI and graphics pipelines. The framework remains Apache 2.0-licensed, supporting Java game deployment on desktop, mobile, and web.
The catch: Active development relies on a small core team, with 333 open issues suggesting maintenance pressure despite steady community use.

Use Cases
  • Build 2D mobile games with Java using one codebase
  • Deploy 3. Create cross-platform HTML5 games with OpenGL ES rendering
  • Prototype 3D simulations for desktop and Android devices

Source: libgdx/libgdx — based on the README and release notes.

Dear ImGui v1.92.8 sharpens tooling workflows for real-time 3D apps 🔗

Latest patch refines widget behavior and backend stability without breaking changes

ocornut/imgui · C++ · 74.9k stars Est. 2014

The v1.92.

8 release of Dear ImGui focuses on polish, fixing subtle input handling quirks in multi-viewport setups and improving font atlas management for embedded targets. Developers integrating the library into game engines or industrial HUDs will notice smoother tab navigation and more reliable clipboard sync across platforms. No API changes mean existing toolchains update safely. The project maintains its zero-dependency, renderer-agnostic core while addressing long-standing edge cases reported in open issues. The catch: Internationalization and accessibility features remain unsupported, limiting use in end-user-facing applications requiring global text handling or screen reader compatibility.

Previously in The Times “covered” — Jul 16

Use Cases
  • Game studios debugging render pipelines with in-engine overlays
  • Embedded systems teams building device configuration UIs
  • Simulation creators tuning parameters in real-time visualizers

Source: ocornut/imgui — based on the README and release notes.

Quick Hits

OpenSpeedy An open-source tool that lets players adjust game speed in real time, enabling precise testing, debugging, and performance tuning of game mechanics without recompilation. 16.6k
renodx A DirectX shader renovation engine that enhances visual fidelity and performance in games by modernizing legacy rendering pipelines with minimal code changes. 1.5k
entt A blazing-fast, header-only C++ ECS library empowering developers to build scalable, cache-friendly game architectures with zero runtime overhead and intuitive APIs. 12.9k
tracy A lightweight, real-time frame profiler for C++ games that captures CPU/GPU timing, memory usage, and draw calls to instantly identify performance bottlenecks. 16.4k
Revelation A visually rich, exploratory Minecraft shaderpack that transforms blocky worlds with dynamic lighting, volumetric effects, and immersive atmospheres while maintaining playable performance. 593
The Git Times AI Desk
Ask about today's stories — or hit “Ask about this” on any article to focus on one.

Unlock the Git Times AI desk to ask about today's stories and the AI model market.

Upgrade to Premium
Answers by the Git Times AI desk · verify before you ship