Fetching from the wire…
Research2026-09-16 · source-backed
GPUThor reverse-engineers GPU memory-access coalescing to build patterns that activate aggressor rows harder than decoy rows, and identifies refresh instances to construct longer patterns escaping in-DRAM mitigations across refresh intervals. Prior GPU Rowhammer work triggered tens to hundreds of flips; this yields 500x to 23,500x more across A4000, A4500, A5000 and A6000 cards, approaching CPU Rowhammer rates. That closes the practicality gap, which matters directly if you rent shared GPU capacity.
Each link below shares sources, entities, or timing with this story.
Everyone writing SKILL.md files has absorbed the same folklore. Keep the top file thin. Push detail into reference files. Let the agent walk the tree as needed. More layers, more context efficiency. A controlled study submitted July 20 tested that across InfiniteBench, three a...
A paper from Xiao Yu, Baolin Peng, and Ruize Xu makes a claim that seems obvious once stated and is genuinely new as a training methodology: modern agents are inseparable from their inference harnesses, so training them in stripped-down RL sandboxes produces a train/serve mism...
Warp released its client codebase under AGPL-3.0, surged to 56,000 GitHub stars and #2 on GitHub Trending. But the real story isn't the open-sourcing. It's the repositioning. Warp isn't calling itself a terminal anymore. It's an "agentic development environment." The product n...
An agent-harness optimization layer offering skills, instincts, memory and security across Claude Code, Codex, OpenCode and Cursor, pushed today, with its last tag v2.1.0 from July 27 (GitHub). If you pin dependencies, the installable surface is a month stale even though the d...
affaan-m/ECC (36.3k forks, MIT) bundles 67 agents, 284 skills, 94 legacy command shims, and "instincts", patterns learned from prior sessions with confidence scores that auto-recall when relevant, plus a .ecc/memory/ markdown vault that's explicitly cross-harness, so context s...
A study of 3,171 GitHub repositories (2,660 multi-component setups, 511 skill collections) measured only byte-decidable defects in Claude Code, Cursor, Copilot and Codex configuration artifacts, validating every finding through independent re-derivation, an LLM adjudicator and...
MindPattern daily
One email a day at 7 AM. Sources and a take on every story. Unsubscribe anytime.