Full-text search across 68431 articles. Combine with topic and date filters; results sorted by relevance.
Ari Wilson built and launched Ari's Reader, a feed reader that uses Claude Haiku to score each new item against interests written in plain English, in about 100 prompts and roughly $206 of Claude Opus tokens at API list …
CISA published advisory ICSA-26-300-01 on October 27, 2026, flagging four vulnerabilities in EPICS Base version 7.0.10 and earlier, including one critical flaw that could allow remote code execution on systems controllin…
A developer described a pattern in which coding agents deliver fixes as prose explanations in chat threads rather than as pushed branches and pull requests, forcing reviewers to hand-reconstruct the change in an editor. …
A paper from researchers at the MIT-IBM Computing Research Lab, with Polina Golland and Yoon Kim of MIT CSAIL, reproduced two self-distillation methods — Self-Distillation Policy Optimization (SDPO) and On-Policy Self-Di…
Researchers at Harvard University and FAIR at Meta released WeightBridge, an open-source weight-transfer library for LLM reinforcement learning that cuts average GPU stall time per weight update by up to 42× compared wit…
A developer, in a talk at SailsConf 2026, outlined an approval-based architecture for AI agents in which the application, not the model, enforces permission checks so agents prepare actions and humans approve only the co…
FreeBSD co-founder Jordan Hubbard released Daemonarchy 1.3.1, a collection of FreeBSD ports that brings the Wayland-based Omarchy desktop to FreeBSD, after roughly 10 releases in about 10 days. The independent effort, un…
Ai2 reported on October 27, 2026 that prompt wording alone moved GPT-5.4's accuracy at judging research-idea novelty from 93.5 percent to 40.9 percent, below chance, when a prompt dropped the note that reviewers found ex…
OpenMoss released MOSS Image, a 5-billion-parameter text-to-image diffusion transformer trained from scratch and published under an Apache 2.0 license covering both weights and code. MOSS Image pairs its diffusion transf…
Physical Superintelligence PBC announced on October 23, 2026 that its researchers, working with its AI research system, obtained new results on five problems from the Open Quantum Problems list, publishing each with a pa…
At Graph Hacks, where 7,899 developers signed up, 647 teams built agent memory systems on FalkorDB in four days in the Agent Memory and Coordination track, and the strongest projects shared five patterns: linking every f…
A new book presents a literate-programming implementation of double-entry bookkeeping in C, building libledger, a library that stores books in a single SQLite file, alongside a ledger command-line tool and an MCP (Model …
OpenAI launched GPT-6.1 Terra on Monday, a computer-use model priced at $1 per million input tokens, $0.05 per million cached input tokens and $5 per million output tokens — half the API price of its predecessor GPT-5.6 …
UsageAtlas, a free and open-source desktop app, reads the local database OpenCode already stores on a user's machine to display tokens, requests, models, sessions and an estimated cost for Windows, macOS and Linux. Usage…
GitHub replaced Copilot premium requests with AI Credits on June 1, 2026, for every monthly plan, leaving only Copilot Pro and Pro+ subscribers on annual plans billed in premium requests until those plans expire. Under t…
UsageAtlas, a free and open-source desktop app, tracks Codex usage by connecting to the Codex CLI already signed in on a developer's machine, showing session and weekly windows, remaining capacity, reset times, tokens, r…
Whytap, an open-source productivity toolkit that runs in the Mac's notch, has been released under the MIT license for macOS 14.2 or newer on Apple Silicon and Intel, with voice dictation, an AI agent, meeting notes and a…
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, 1,209 calls in 34 sessio…
A coding agent working on the DynoInvoice e-invoicing system under Keel wrote 17 negative tests in a shared Kotlin base class, but 16 of the 17 never executed because the test methods used expression bodies that returned…
A hands-on evaluation of the codebase-memory-mcp MCP server found that the tool's installer writes to settings.json and installs PreToolUse, SessionStart and SubagentStart hooks across 43 auto-detected coding-agent clien…