Developer Corner

An open, vendor-neutral tester checks AI chips for wrong results that may appear without a crash or hardware error, and publishes evidence in OCP Test and Validation format.

A hands-on telecom lab presenting 4G and 5G core networks, roaming, IMS calls, and packet-level scenarios.

An open-source library for driving Android phones from Python, Node.js, the command line, HTTP, or MCP.

A technical showcase for the Veri 1.20B language model, with an interactive architecture explorer and reported training and optimizer comparisons.

An evidence-led investigation compares Armada’s privacy and hosting claims with its app, infrastructure, policies, and public materials.

A source-linked collection of reports about unexpected Cloudflare bills, their reported causes, and what happened afterward.

A research microsite describing an LLM backdoor that uses differences in inference hardware and software as its trigger.

Lithance says it turns a plain-language hardware brief into KiCad files and manufacturing deliverables, with automated checks attached.

A transit-data service that lets compatible AI assistants look up arrivals, stops, routes, and vehicle locations across Ontario agencies.

An early-stage rail R&D project describes models it hopes to build to predict switch failures from wear and other trackside signals.

A triage console for security researchers to assess leaked-credential findings offline and prepare responsible disclosure drafts without including the secret itself.

Shotcup offers deterministic, short-lived JavaScript sandboxes for AI agents, with runs available through REST and MCP.

Algovista is building GPU-accelerated virtualization tooling to run Windows 7 and Vista, including Aero glass, on Apple Silicon Macs.

Cordon offers an open-source scanner for dependencies, repositories, pipelines, MCP servers, and infrastructure, with a cloud product for findings and policy.

This open-source CLI guides an AI coding agent through removing a feature flag and its dead code, then checks the build, tests, and remaining references.

A suite of MCP servers, CLIs, and document tools for connecting AI assistants to Korean law, official documents, and government data.

A searchable reference for offensive-security commands, with command-building, MITRE ATT&CK coverage, and detection-engineering material.

It connects through MCP and lets an agent inspect code, data, pipelines, jobs, and infrastructure as one map.

Ahope offers micro- and nano-device prototyping, from design and process planning through fabrication, as well as individual cleanroom processes.

Altifield describes a native macOS canvas for arranging CLI agents, terminals, browsers, and automations in one workspace.

chroot.dev offers OpenBSD virtual machines in Firecracker microVMs, managed through SSH commands and private HTTPS endpoints.

CloudZH is a paid dashboard for installing and managing media and server apps on a Linux machine, including Plex, Sonarr, Radarr and WireGuard.

A Python package designed to add business context and read-only database querying between an AI agent and company data.

An open-source macOS tool runs coding agents in separate repository forks, tests their patches, and keeps the first one that passes.

Hostkeep connects an AI assistant to Windows machines and VMs to inspect logs, code, and outputs, and run explicitly allowed tasks.

Husk is an open-source tool that gives AI chats a terminal, files and a browser, with results kept in workspaces you can reopen.

A free, open-source macOS terminal IDE combining command history, a code editor, Git review, and support for coding agents run in its terminals.

A frontend QA command-line tool that checks routes in real browsers across user states and reports what it did—and did not—measure.

Perch is a command-line linter that lets teams write project rules in plain English and run them alongside code scans.

adhdTerminal is a $99 Mac app that keeps multiple Claude Code sessions visible, labeled, and resumable in one window.

A local tool routes questions from coding agents to a human, then sends the answer back through the agent’s running session.

A robotics physics engine in development, aimed at simulating materials such as cloth, cables, bags, and other soft goods from real-sample measurements.

DevTerminal is a macOS terminal that detects project types and maps ⌘R to their run or build workflow.

Nitpick lets reviewers mark a spot on a webpage and attach a note, then packages it with page and browser context for a development team.

A small software lab presents MyCoder, a policy-enforced terminal coding agent, alongside local-first research and exam-review tools.

A local CLI reads coding-agent histories and makes a shareable report with a profile card, usage counts, and cost summary.

A private-beta tool that says it tracks API and SDK changes, traces their impact through code, and opens pull requests with fixes.

A tiny decision API offers fixed yes-or-no answers, emphatic refusals, and a Magic 8 Ball endpoint.

A hosted mock SMS API for testing OTP flows without sending real texts.

A macOS tool that turns Claude Code or Codex sessions in Neovim into spoken, interactive pairing sessions.

RoboPay demonstrates a Solana escrow flow that releases payment to a delivery robot’s owner after the robot scans a parcel’s QR code.

A Linux command-line tool presented for finding sensitive configuration files, flagging risky exposure, and creating encrypted backups.

An open-source desktop app for running coding agents side by side, with an isolated checkout for each and a shared queue for prompts, crashes, and finished work.

It connects TallyPrime accounting data to HTTPS endpoints for reports and writes, with an MCP server for AI assistants and other clients.

An independent research lab cataloguing work on long-running coding agents, efficient inference, and model diagnostics.

A local retrieval tool that packages document chunks, vectors, and an embedder into a .tome file for querying alongside a locally run model.

Crewboard describes a shared workspace where a team can assign work to coding agents running on each member’s own computer, with completed tasks returned as pull requests.

Wardian presents a local-first environment for making, running, and sharing small utility apps, including apps created with AI.

CompsAPI offers a JSON API for eBay sold listings, with search, pricing comparisons, and filters for trading-card sales.

CSJ TechLab reconstructs refinery and petrochemical piping drawings as 3D models, linking equipment to P&IDs and related information.

IP Almanac offers reverse-IP lookups and DNS history, with network classifications and observation dates for domains and IP addresses.

A memory plugin saves selected coding-agent sessions so Claude Code, Codex, Cursor, and Pi can search earlier decisions and fixes.

A Catalan-language manual describes a CLI and terminal interface for capturing and analyzing network traffic, including distributed capture.

Patina Source offers game studios payments to license finished, cancelled, or unreleased games’ source code as AI training data.

TerminaTR is a terminal-based coordinator for coding agents, assigning work to separate threads in Git worktrees and tracking them through a persistent dashboard.

TestXector converts and inspects STIL, WGL, VCD and eVCD patterns for Advantest and Teradyne workflows, with migration assessment and comparison evidence.

An open-source desktop tool for turning natural-language requests into Roblox scripts, models, interfaces, and game changes inside Studio.

Vyrtel is an open-source, self-hosted tool for collecting and exploring logs, traces, and metrics.

TICK offers a client that places a sponsored text line beside supported AI coding tools while a model is working, sharing ad revenue with users in crypto.

A one-person Michigan engineering lab reverse-engineers mouse and keyboard firmware, builds accessible controllers and drivers, and takes on software consulting.

OpenAgent adds a voice interface and local desktop controls to Instinct, including shell commands, code edits, native Mac apps, authenticated Chrome, and social accounts.

An open-source terminal agent for authorized security testing that maps a target, tests security hypotheses, and writes up findings.

Unhappen describes a runtime layer that routes AI-agent tool calls through snapshots, rollback, and policy checks.

A secrets manager for AI coding agents that presents secret names to the model and injects values into child processes.

A Windows utility that watches server login failures and adds repeat-offending IP addresses to Windows Firewall block rules.

An early-stage Claude Code plugin for chip physical design that looks up and generates Cadence Innovus Tcl and guides engineers through a structured backend flow.

Heel presents a prototype vault policy for limiting what an AI agent can spend through Solana.

A Jakarta-based systems and integration engineer presents tools for analyzing SMS gateway logs, working with ISO 8583 messages, and inventorying OpenEdge ABL code.

An interactive prototype imagines websites keeping persistent identities as authorized machines take turns hosting their files.

A testing tool that runs an AI agent, introduces disruptions, and reports how it responds.

A source-backed claim-verification endpoint for autonomous AI agents, listed at 0.20 USDC per verification on Base Mainnet.

A service for product manufacturers that receives vulnerability reports, escalates urgent cases, and helps organize Cyber Resilience Act reporting.

PBI proposes a bond market for APIs, MCP servers, and data feeds: operators stake capital against measurable service promises, with failures triggering formula-based settlement.

postcache describes an API that gives AI agents temporary public media URLs for Instagram, Threads, and Facebook posts, with platform checks and file expiry.

Proksolab presents proximity-sensing modules, evaluation kits, integration support, and custom hardware for product makers.

A command-line tool for testing row-level security across configured Power BI and Microsoft Fabric personas, with failed checks able to stop a CI step.

A browser-based tool and JavaScript harness for seating AI agents at a question, moderating their turns, and producing minutes.

TriageKit presents a macOS menu-bar app that checks iOS Simulator screens for visual bugs, traces findings to source files, and prepares patches for approval.

MyTrackables checks whether a field in a public HTTPS JSON record has the value you expect, then provides a signed evidence receipt.

Quell presents itself as a proxy that classifies AI-agent database and cloud actions, then allows, holds, or blocks them.

Kinora is an open-source, self-hosted streaming server with a web interface and an Android TV app. It searches its configured sources when a viewer presses play.

InQ presents an SDK for adding QR check-in, live waitlists, and turn alerts to point-of-sale systems and apps.

A proposed handoff layer for sending already-verified carbon data to regulators in each body's required format.