The GAL (General Agent Loop & Executor) project released a single-binary AI agent runtime that runs inside a kernel-level OS sandbox by default across Linux, macOS, and Windows. The runtime ships CLI …
A developer built a public "oracle" application around Perplexity's pplx-decider-flash-v1.2, a 4B-parameter decision model with its language head removed that returns calibrated probabilities over 255…
A developer released Commex, an open-source, self-hostable alternative to Disqus that lets site owners keep their own comment data. Users create an account, configure an organization with their own Mo…
Patrick, founder of Pidgeon, built an email API that lets AI agents obtain their own persistent addresses and wait for verification codes without polling. The await endpoint holds a request open for u…
Naetive has launched a tool that gives multiple AI coding agents a single shared understanding of a project, connecting Claude Code, Codex CLI, ChatGPT, Claude on the web and Claude Desktop through th…
Pang Jun Han, founder of MyCreatorPulse, built an AI workflow for a Malaysian construction company that reads signed delivery orders, matches them to purchase orders, and flags missing paperwork. He s…
An open-source Python CLI called git-translate-commits translates Git commit messages into a target language, defaulting to a local Argos Translate engine that needs no API key. The tool creates a bac…
A developer at Appwrite analyzed 1,929 Playwright MCP browser calls across 62 Claude Code sessions from 17 September to 17 October and found that almost none ran the existing test suite; instead, 1,20…
A developer documented three flaws in per-user and per-IP free-tier caps placed in front of paid LLM calls on a Lambda-behind-CloudFront-and-API-Gateway stack: keying on X-Forwarded-For or a client-su…
A developer tracked seven months and $28,335 of coding-agent usage across Claude Code and Codex, logging 1,432 sessions and roughly 200,000 tool calls, and found that only 0.7% of tool calls occurred …
The MCP working group released protocol revision 2026-07-28, described as the largest revision since launch, which removes the initialize handshake, session IDs, and held-open streams in favor of stat…
A developer outlined a layered architecture for defending AI agents against prompt injection, arguing that prompt engineering alone cannot stop attackers from hijacking an agent's instructions. The ap…
An engineer described a method for correcting an LLM judge's reported pass rate using roughly 100 human-labeled outputs, applying the Rogan-Gladen correction borrowed from medical testing. The approac…
An open-source project called Scout provides a browser-using AI agent that completes web tasks from plain-language instructions, reading pages, clicking, typing and scrolling until finished. On Outcra…
A developer published a five-stage tutorial for building a token ledger that tracks tokens per accepted output rather than tokens per call, arguing the former is the metric that determines whether fre…
A developer argues that cancellation in full-stack AI generation features should be modeled as a first-class terminal state in a distributed transaction rather than an exception path, citing a support…
A developer published a stdlib-only Python script, lane_fit.py, that filters AI coding lanes by hard constraints—data class, concurrency, model pinning, audit logs, and ops ownership—before any price …
A developer outlined a workflow for using AI and IoT to flag construction schedule risks before they surface in project plans, starting with a single measurable condition such as an activity about to …
A developer built a motion-design pipeline in which Claude Opus 5.5 writes a single self-contained HTML file where every frame is a pure function of time, Playwright screenshots each frame, and ffmpeg…
The developer behind Kloner, an AI website cloning tool, outlined a pipeline that renders pages in a headless browser, captures the final DOM and computed styles, and rebuilds the result as editable, …