A developer documented how to read the fsType embedding-permission field in the OS/2 table of TrueType and OpenType fonts using fontTools, showing that the low four bits hold a single usage value (0, …
A developer built PressBridge, a self-hosted MCP server that lets Claude manage WordPress sites over SSH without exposing wp-config.php, the database password, or other sensitive files. Every permissi…
A developer built Séance, an offline chatbot that uses the open-weight Gemma model via Ollama to role-play the "spirits" of seven dead programming languages including COBOL, FORTRAN, ALGOL, LISP, Smal…
Engineer Naveed Ahmed outlined an architecture for giving AI agents limited access to production infrastructure, using the Model Context Protocol (MCP) to expose narrowly scoped diagnostic tools rathe…
A developer built Exorcist, an open-source dependency health scanner for package.json that runs entirely in the browser and flags issues npm audit misses, including deprecated packages, functional dup…
A developer ported a PII tracer byte-exact from Python to TypeScript so it can run inside a coding agent's request pipeline rather than alongside it, aiming to strip secrets such as API keys before re…
Anthropic released Claude Haiku 5.5 (model ID claude-haiku-5-5) on 7 October 2026 with a comparison table against Haiku 4.5, GPT-6 Luna and Sonnet 5.5. According to Anthropic's published figures, the …
A developer has proposed making AI agents a first-class principal in IAM, built on the rule that an agent owns no access and instead borrows it from the user it works for, one request at a time, with …
A developer built a set of 45 open-source AI tools running entirely on local models via Ollama, avoiding paid API calls and sending no data to third-party servers. The stack uses llama3.2 and other sm…
A developer published an SVDQuant 4-bit quantization of the Qwen-Image-2.1 Turbo image model and its Qwen3-VL-8B text encoder, cutting text-to-image peak VRAM from 26.44 GiB to 17.23 GiB and enabling …
A developer published a deep-dive arguing that prompting accounts for only about 20% of a working AI feature, with the remaining 80% made up of orchestration, tool reliability, context management, gua…
Powerduck published a guide showing developers how to generate an MCP server from an OpenAPI spec, run it locally, and register it in Cursor's MCP configuration so the coding agent calls typed tools t…
A developer describes an intent-log pattern for preventing AI agents from duplicating writes to legacy systems of record. The approach routes agent tool calls through a wrapper that writes a normalize…
A developer has introduced SocialSync Ghost Node, a static, client-side dashboard for the Hacktoberfest Open-Source AI Challenge that highlights which concepts an open-weight model such as Google's Ge…
A developer argues that AI agents executing real actions inherit classic time-of-check-to-time-of-use (TOCTOU) bugs from distributed systems, illustrating how an agent that verifies a refundable balan…
A developer measured the practical limits of AI-generated PNG logos, finding that a 1,200-pixel PNG stretched across a one-metre banner renders at roughly 30 pixels per inch, far below the 300 dpi pri…
A developer running LLM workflows in n8n in production described how an OAuth token and its rotating refresh token expired silently, causing a CRM enrichment workflow to stop writing data while every …
A developer published a practical guide to sizing the KV cache for local LLM inference, showing that cache memory scales with layer count and KV head count rather than parameter count. Using the formu…
Security researchers Gal Weizman and Omer Samet of Forever Security demonstrated that a web page can silently plant an AGENTS.md file in a user's Downloads folder, which OpenAI's Codex coding agent th…
TuringGate Lab launched a weekly AI-jargon series, Suggested Jargon, publishing AI-written and AI-disclosed posts that rank terms by mention counts from a daily scan of Hacker News headline feeds and …