Recently Evaluated

A research write-up on detecting AI agents through process differences in CAPTCHA and related cognitive tasks. It outlines the CogCAPTCHA30 approach, reports human-vs-model differences, and connects the findings to Roundtable’s Proof of Human product.

A lightweight, single-binary terminal coding agent harness written in Go. It supports multiple model providers, local OpenAI-compatible models, built-in file and shell tools, extensions, session management, side chat, and swarm subagents.

Open-source C++ and CUDA LLM inference engine inspired by vLLM, with a teaching-focused course that walks through model serving, batching, KV cache, and attention kernels.

arXiv paper on a self-speculative decoding framework for speeding up reasoning LLM inference on edge hardware, with hardware co-design and reported speedups.

A practical guide to building an always-on personal AI assistant with Pi, OpenCode Go, Telegram, Resend, Hetzner, Tailscale, GitHub, and Nginx.

AppFactor’s blog post explains an AI-powered MCP Bridge feature that makes legacy API tools easier for agents to find and use by combining keyword search, vector search, and AI-generated enrichment from schemas and sample responses.

A CLI tool that replays Claude Code session logs to reconstruct lost project files commit by commit. It includes deterministic replay from JSONL logs, an optional Claude-based classifier for Bash events, and install/run instructions.

A Windows desktop agent that lets AI interact with any app through visual GUI control, with a portable single-EXE build and support for ChatGPT, Gemini, and Claude.

A GitHub repository for a latency-separated AI memory retrieval and RAG system. The README describes fetch, compute, and ANN search stages, includes benchmark ranges, and exposes a public test endpoint.

Chrome MV3 extension that organizes existing AI tabs into a structured model council for cross-checking prompts across ChatGPT, Claude, Gemini, Perplexity, Merlin, and Grok.

Repolog scans a live website and produces a ranked audit covering on-page SEO, Core Web Vitals, security checks, and AI readiness for major AI search and assistant platforms.

OpenHive is a shared knowledge base for AI agents to search and post problem-solution pairs. The page shows multiple integration paths, including Agent Skills, MCP, REST API, and discovery endpoints.

CLI for catching AI-generated code smells and regressions in code. It scans changes with 40+ rules across 7 languages, offers fixes, CI gating, hooks, and MCP tools.

Open-source context layer for data and analytics agents that adds warehouse context, semantic search, and MCP/CLI access for tools like Claude Code and Codex.

A Claude Code stop hook that detects negative feedback in the conversation transcript and plays a random local sound from a folder, with a fallback voice prompt if no sound files are present.

A TypeScript static scanner that finds AI agent tool calls with no checks before they can trigger real-world side effects. It supports CLI scanning, JSON output, a generated tool-call registry, and CI/pre-commit enforcement.

DeepSWE is a benchmark for measuring frontier coding agents on original, long-horizon software engineering tasks. The page shows a leaderboard, methodology overview, task examples, and a full blog explaining the benchmark design and results.

arXiv paper describing AVA, a GenAI platform for policy and development research built on 4,000+ World Bank reports. The abstract highlights multilingual support, evidence-based synthesis, citation verifiability, and reasoned abstention when queries cannot be supported.

A Claude skill for generating better commit messages with gitmoji + Conventional Commits, atomic commit enforcement, and a hook guard to keep commits on-format.

A Superconductor blog post showing how background coding agents were used to reproduce, diagnose, and fix a Rails memory leak using derailed_benchmarks, with a reusable Agent Skill workflow included.

Enough is a beta personal language system for planning, writing, reviewing, and translation. It supports local models and OpenRouter, and is aimed at users who want more control over their data while building a flexible personal knowledge workflow.

Agent skills for VillageSQL that help AI coding tools build a VillageSQL extension, including a 7-phase workflow for the vsql-extension-builder skill and install support for Claude Code, Gemini CLI, Codex, Cursor, Amp, and Kiro.

Dream Server is an open-source stack that turns a PC, Mac, or Linux machine into a private AI server with local model inference, chat UI, voice, agents, workflows, RAG, and image generation. The page shows setup instructions, platform support, release notes, and a clear positioning for self-hosted use.

Multiplayer is a local debugging agent for developers that connects coding agents to production data so they can fix application bugs automatically. The site highlights runtime debugging, issue deduplication, and integrations with tools like Claude Code, Codex, and Copilot.