Developer Corner

An unofficial, free Windows 10/11 port of SSX Tricky, built from the Xbox game and requiring players to bring their own copy.

Pointfix connects an iOS simulator, Android app, or web page to a coding agent so developers can select an interface element, describe a change, and review the result.

Recomputable is a pilot network for sharing research compute, with jobs designed so other machines can rerun them and check the results.

A writing-analysis API checks text for specific clichés and style habits, returning a probability for each factor.

An open-source command-line tool that drives long, narrow USB displays from macOS, Linux, or Windows.

An open-source macOS utility mirrors iPhones, iPads, and Android phones, then lets you interact with them using a Mac’s mouse, trackpad, and keyboard.

An early-stage robotics project exploring whether small machines can prepare, place, and compact soil into rammed-earth walls.

REA helps coding agents inspect software and explain how its behavior works, with guided examples and setup instructions.

A Base Sepolia test project gives an ordinary Safe a passkey-backed owner seat that uses a one-time key for each approval.

Deck is a free, open-source code-review tool where engineers can inspect an AI agent’s decisions against the changed code, question them, and choose what ships.

A commercial audit service tests code-training environments for cases where a grader awards full reward to an incorrect fix, then supplies replacement tests.

A free, open-source macOS terminal designed to organize coding agents, their output, and related browser work.

FLYDER describes a simulation in which a fruit-fly brain model steers a spider-like crawler through real web pages, with clicked-link words used to name a pump.fun coin.

A local CLI runs fixed behavioral probes against MCP servers and grades each test pack with supporting evidence.

The site documents an AI agent’s claimed run buying a domain, hosting, a repository, and media through pay-per-call services.

An interactive heatmap charts projects posted across nine builder subreddits, comparing each theme with its own 12-week average.

A security-testing workflow that records scoped checks as hash-chained evidence, then produces reviewer-signed reports that can be verified offline.

Bifurc is a desktop app and CLI for routing, capturing, mocking, and testing APIs running on your machine.

An open-source desktop, command-line, and Python suite for building, simulating, and analyzing polymers and materials.

A browser-based workspace for comparing OpenAPI JSON versions and preparing a release checklist.

Driftfix says it opens tested pull requests in customer repositories to migrate code away from an API’s older version.

A research project documenting the electronics, software, and testing work for the CMS electromagnetic calorimeter upgrade ahead of the High-Luminosity LHC.

A free macOS app that bundles PHP development essentials—including MySQL, local HTTPS, Node and a mail catcher—in one menu-bar environment.

Jokku presents a self-hosted deployment platform for building apps from a Dockerfile and running them across servers you own.

A browser-based research preview demonstrates agent-to-agent message sealing with ML-KEM-768 and lets visitors inspect and tamper with the resulting envelope.

Nox is presented as a statically typed language with Python-like syntax that compiles ahead of time to native code.

An open-source CLI turns AI coding-agent sessions into shareable receipts and can block risky commands or request approval.

A public benchmark compares decision models, chat models, and simple programmed strategies as they guide an agent through grid puzzles.

AgentOrbit gathers questions and completed work from coding agents into an inbox you can answer from a Mac or iPhone.

Assay tracks an AI-assisted SQL engine, separating benchmark progress from machine-checked proofs that its answers are correct.

A draft protocol proposes that neighbors resell network capacity hop by hop, with payment receipts serving as proof that data arrived.

A Mac app presents Claude Code, Codex, Gemini, and OpenCode sessions as contacts in a retro-style messenger.

Cutscene is a private-beta tool for asking a coding agent to turn an app into a short, code-rendered launch video.

An outside-in scanner checks a website for publicly reachable environment files, backups, configuration files, logs, and other sensitive artifacts.

A desktop engineering tool for building or importing mechanisms and vehicles, running simulations, and inspecting results in plots and 3D playback.

EMT Pilot says it can predict, run, and document PRC-029 ride-through studies for solar, wind, and battery plants using an engineer’s existing EMT simulator and OEM models.

FlyOps describes a system for assembling temporary illuminated structures from smart LED cubes using autonomous drones.

Glimdock is an open-source device-monitoring system that gathers readings through a local Rust collector and shows them in a browser and on a small desk display.

A Model Context Protocol server offers coding agents tools for domains, DNS, deploys, secrets, storage, email, messaging, and pre-launch checks.

A set of free calculators and reference tables for estimating LLM memory, GPU costs, fine-tuning, and self-hosting versus API use.

An open-source macOS workbench for inspecting authorized Mach-O, PE, and ELF files locally.

An MIT-licensed tool for coordinating multiple coding agents on one Git repository, with separate worktrees, a shared status console, and a gate before changes reach the main branch.

Labeltap presents a way for AI agents to find and operate Android app controls by name rather than by screen coordinates.

Patchkite presents a self-hosted tool for sending over-the-air updates to React Native and Flutter apps without waiting for a store release.

Pincite presents a schematic-review tool that checks circuit designs against component datasheets and attaches quoted passages and page numbers to its findings.

A 3D modeling studio built around procedural models that can be adjusted with parameters and an AI assistant, then used in Unreal, Unity, or Blender.

Reams is a self-hosted collaborative wiki whose MCP server lets agents search, read, and edit team documentation.

An open-source fan platform for playing skate. offline, hosting multiplayer sessions, and installing community mods.

Sturdybench offers test cases for checking whether an AI agent or MCP server chooses the right tool and arguments—or correctly makes no call.

A comparison tool estimates bills for 237 infrastructure and API services at a usage level you choose.

A macOS safety layer that checks Claude Code tool calls against local rules, sends ambiguous cases for contextual analysis, and can pause high-risk actions for human review.

A local-first desktop proxy lets developers switch individual services between a machine, QA, or a mock while capturing traffic.

A waitlisted Mac app designed to monitor several coding agents and let developers respond to them from the notch.

An open benchmark measures browser-based GPU performance and estimates which language models a machine may be able to run.

An MCP email server for agents to add domains, create mailboxes, read a shared inbox, and send or reply to messages.

An early-stage copilot that compares a reference data pipeline’s streaming and batch paths, then explains discrepancies using pipeline evidence.

A Wisconsin-based independent builder offers pay-per-result JSON APIs for weather, forecast markets, and other public and business data.

Framegaze describes a local-first MCP server that helps language models inspect video frames, audio, and speech on a shared timeline.

An open-standard draft with an in-page editor and browser-extension workflow for setting rules about how websites, apps, and AI agents use data and interact.

Synaptic OT presents an on-premises security platform for monitoring industrial networks, investigating incidents, and preparing human-approved responses.

Tuvio says it scans code repositories for exposed database access, leaked secrets, vulnerable packages, and other security mistakes.

VMAFx documents a toolkit for perceptual video-quality assessment, including its metrics, interfaces, and hardware backends.

Breakspec is a proposed testing tool for AI-written code: it checks approved rules against inputs the code’s own tests may have missed.

HexDB presents itself as an open-source Rust document database combining JSON APIs, search, transactions, encryption, and replication in a single program.

Prism DLP describes a workplace tool that checks text and files sent to AI services, then lets them through, redacts sensitive details, or blocks the transfer.

Blueberry Jam presents Codify, middleware that says it learns the scope of recurring agent tasks from observed runs and enforces that scope at the container boundary.

Nock is a CLI tool for collecting completion notices and approval requests from AI agents and other commands, with remote control and audit features.

KeyValet describes an open-source macOS credential broker that lets AI agents use API keys and OAuth accounts through MCP without seeing the credentials.

A coming-soon Mac-and-iPhone companion pitched as a way to approve coding-agent prompts and supervise sessions remotely.

A certificate-management service that says it can issue, renew, monitor, and deploy SSL certificates for your domains.

An early-access project plans a panchang API that calculates Hindu calendar dates for a user's location and timezone, with the reasoning attached.

A Rust library applies centering and scaling to matrices without materializing a separate normalized matrix.

The Ukrainian engineering company presents trailer platforms, mission modules and sensor systems designed to work with ground robotic platforms.

A bilingual threat-observation dashboard that says it collects scans, vulnerability probes and credential attacks, then organizes activity into maps, timelines and classifications.

A local reverse-engineering project that coordinates specialist agents and gathers their findings into hash-stamped evidence bundles.