AI · 135 POSTS · PAGE 5/14
AI.
Writing tagged AI: 135 posts on AI systems, engineering tradeoffs, and building products that have to work in production.
A similar skill took 1 in 5 runs from the installed one in 6,368 Claude Code runs, with no drop in task completion. A pre-tool hook at the first skill read fixed most of it.
Across 2,124 runs of Codex, Claude Code and Kimi Code, letting the lead agent spawn sub-agents raised token use 1.41–3.31× and cut SWE-bench Verified pass rates for two of three agents.
ParanoiaEval measures unnecessary defensive work by coding agents: 11.2%–58.7% of runs despite evidence against it, hidden behind 88–98% task success. Bounds written into instructions help most.
47.3% of conflicts in 715 agent PR pairs involved disjoint authored files; a lineage check beat file overlap (0.877 vs 0.694 AUROC), and replay costs 0.02 s.
Across 1,000 runs of three coding agents, declared dependencies matched runtime reality poorly and never repeated across trials. Add a manifest gate next to your tests.
ASAD sizes its agent team to bug difficulty and beat both fixed and maximum teams on medium bugs while using far fewer tokens. Agent count should be a decision, not a default.
CONTRA keeps a clarifying question only if code written under two different answers behaves differently, reaching 41.2% F1 vs 27.3% for the best baseline. Downstream pass@1 gains are about a point.
Review quality tracked whether evidence was grounded in execution, not reviewer size, but the deployable cascade still rejected 66% of good patches.
HiSentinel vets each agent action before it runs and lifted Qwen3-Coder from 30% to 44% on SWE-bench Verified Mini. Builders get a cheap pre-execution hook idea, with a small eval.
Across five-turn task chains, agents' repo exploration fell from 76–91% to 16–51% and re-implementations climbed past half, while pass rates held. Interface summaries doubled reuse; full source did not.