Fetching from the wire…
Skills2026-08-07 · source-backed
EA-Graph stores verification claims tied to specific artifacts, tracks evidence strength separately from freshness, and marks a claim unprovable rather than fabricating when the content is gone. Beat prose notes in all seven test environments on Haiku (paired Wilcoxon p=0.0156). Prose notes preserve a conclusion without the state that justified it, which is why re-derivation is expensive.
Each link below shares sources, entities, or timing with this story.
I've spent real hours tuning the CLAUDE.md in my own repos. Rewriting architecture notes. Adding conventions. Trimming when it got long. So this one stung. arXiv 2607.27250 ran a two-agent ablation across Claude Code and Codex: 17 real tasks from 3 repositories, 288 gold-test-...
Tessl ran 880 evaluations across 9 models with and without agent skills. The result inverts what most teams assume about AI costs. Haiku 4.5, Anthropic's cheapest model at roughly $0.25 per million tokens, scored 84.3% when given a well-crafted agent skill. Opus 4.7, the most...
Nemotron-Terminal (arXiv:2602.21193, 66 HF upvotes) — First systematic study of data engineering for terminal/CLI agents. Terminal-Task-Gen pipeline with Dockerized environment interaction. Qwen3-initialized 8B model goes from 2.5% to 13.0% on Terminal-Bench 2.0. All checkpoin...
Two thirds. Not two thirds of a contrived jailbreak set. Two thirds of realistic malicious issue requests, against the exact three tools most of the people reading this run daily. Ankur Singh, Jinqiu Yang, and Tse-Hsun Chen built IssueTrojanBench across four attack categories...
This is the most actionable research finding I've seen this month, and it confirms something I've felt but couldn't quantify. Paper arXiv:2604.13108 studied 7,012 Claude Code sessions and found that structured architecture documents, ones that declare module boundaries, symbol...
If the margin collapse is happening to the providers, the demand-side version is happening in your own usage graph. Practitioners are now stacking independent token-reduction layers and compounding the savings past 90% total. (Codepointer) The recipe that's making the rounds:...
MindPattern daily
One email a day at 7 AM. Sources and a take on every story. Unsubscribe anytime.