Posts

Browser automation agent workflows fail most often not because the AI lacks skill, but because the interfaces it targets were never designed with agent reliability in mind. WorkCanvas Studio is an open-source, synthetic fixture library that gives you two realistic business screens — a Standard HTML inventory inquiry and a React purchase-order form — plus four public workflow plans, so you can evaluate and regression-test agents against the precise controls that break them, without credentials or production data. ...

Barehands is a free, open-source project that turns a standard webcam and the Chrome browser into a glass board you control with your hands — no headset, no controllers, no gloves. It uses Google MediaPipe to track your hands in real time and lets you tap, drag, scale, and throw items across your screen, even wiring in an AI that sees, hears, and acts through the same interface. What Is Barehands and Why It Matters Barehands (jaredrhod/barehands on GitHub) is a webcam-based hand-tracking interface that lets you control a 3D “glass board” on your screen with natural hand gestures. Created in mid-August 2026, it already holds roughly 907 GitHub stars and about 199 forks, according to the GitHub API as of September 7, 2026 — remarkable velocity for a project less than a month old. ...

Legal skills markdown agents are turning legal work into executable playbooks: a public repo called legal-skills packages U.S. utility-patent work as SKILL.md files, checklists, and deterministic Python scripts that run a 316-item pre-filing audit without opening a matter. The claim is blunt — displace legal fees, not assist counsel — and it sits on a broader movement where legislation itself becomes versioned Markdown in git. This guide explains what the pattern is, why Markdown is the right format, where agents still fail, and what stays reserved to human lawyers. ...

MarkMem is a git-native, plain-markdown memory layer for chatbots that stores every fact as an inspectable Markdown file in a git repository, backed by a rebuildable SQLite cache. It combines bi-temporal facts in YAML, a Presidio PII write gate, and git as the compliance surface, so you can cat, grep, and git-diff your chatbot’s memory instead of trusting an opaque vector database. What is MarkMem and why do chatbots forget? Every chatbot that carries context across sessions faces the same problem: the model’s context window is finite, and once a conversation scrolls past it, the details are gone. Traditional memory layers solve this by embedding facts into a vector database and retrieving the most similar chunks at query time. That works, but it turns your chatbot’s memory into a black box — you cannot read it, audit it, or prove what it contains. ...

A GEO AI search optimization agent is a software system that automatically audits your content for AI-search citability, rewrites it to maximize visibility in LLM-generated answers, and verifies that ChatGPT, Perplexity, Gemini, Claude, and Google AI Overviews actually cite you. You build it by assembling open-source building blocks: a 0-100 readiness scorer, an AutoGEO-style rule extractor, a rewrite engine, and a citation verifier wired together over an MCP-compatible loop. This guide walks through each step. ...

A planner-executor agent workflow MCP orchestrator splits a hard task into two roles: a planner that decomposes the goal into a step-by-step plan, and one or more executors that carry out each step with tool access, all connected through Model Context Protocol (MCP) tool servers. This pattern handles open-ended, path-dependent work that a linear pipeline cannot, because the planner can re-plan mid-run based on what executors return. It is the canonical multi-agent orchestration architecture for tasks where the full sequence of steps is not known in advance. ...

Vecline is a bit-exact SVG and document toolkit that converts flat art — logos, icons, UI, screenshots, pixel art — into editable SVG with perfectly measured output: SSIM 1.0000, PSNR infinity, zero differing pixels. Unlike most vectorizers that claim close-enough accuracy, Vecline re-renders every SVG it produces and scores it against the source, and its lossless mode fails rather than returning a near-miss. Built for JavaScript as a zero-dependency core with an MCP server, it is the honest answer to overhyped “100% accuracy” vectorizers. ...

PromptShield is a prompt injection scanner that audits repositories for malicious instructions and hidden Unicode before AI agents ever consume them. It statically scans agent instruction files, detects zero-width character attacks and taint patterns, and maps findings to the OWASP Agentic Top 10 (2026) so teams can block supply-chain and prompt-injection risk in CI/CD. What Is PromptShield and Why Repo Scanning Matters PromptShield belongs to a fast-growing category of tools built to answer one uncomfortable question: can you trust the instructions your AI agent is about to read? As coding agents like Claude Code, Codex, and Cursor become the default way teams ship software, the files those agents read — READMEs, AGENTS.md files, MCP configurations, and skill definitions — have become a new attack surface. ...

LabLLM is a free, native macOS app (Swift/SwiftUI) that lets you train tiny language models from scratch on Apple Silicon — no cloud, no subscription, everything stays on your Mac. Released as a beta in August 2026, it bundles a model builder, dataset browser, tokenizer, training dashboard, LoRA and DPO fine-tuning, a sampler, a chat window, and checkpoints into one GUI, so you can watch a small LLM “emerge” live on your own machine. ...

Apeireth is a pure Safe Rust project that bills itself as an “AGI Operating System” base — a cognitive microkernel with continuous topological memory and a causal world model, built with #![forbid(unsafe_code)] across 16–17 crates. It is extremely early-stage: created September 4, 2026, roughly 27 stars, one fork, and all benchmarks are self-reported. This review separates the ambitious vision from the verifiable reality so you can judge whether it is a genuine foundation or an overpromised prototype. ...