DAIR.AI indexes all 1,760 curated AI papers, now chat with them by topic
omarsar0 · x · 2026-08-27
DAIR.AI (@omarsar0), curating AI papers for 5+ years, has indexed all curated papers in one place — discover by topic and chat with them; affiliations and citations coming soon. The collection totals 1,760 papers across 176 weekly issues since April 2023.
This week's 10 picks include:
- Agent Lightning v1.0 (Microsoft): treats the agent harness/trainer boundary as the integration point for training
- Skill Trigger Bottleneck: 56,804 public agent skills compete for fewer than 100 reliable trigger slots, so the long tail never gets used
- Harness-Level Forgetting: agent continual learning happens in prompts/memories/tools/skills, a layer nobody measures for drift
- Control-Plane Tax: instrumenting 10 real agentic apps shows the model is often not what your latency bill pays for
- Demystifying Agent Skills: controlled comparison finds the "skills inject missing knowledge" assumption almost entirely wrong
Related event: DAIR.AI Launches Indexed Library of 1,760 AI Papers with Chat Feature(2 posts)→
More from coding & agent
- Offline Wikipedia RAG via 100-line Bash script works on local LLMs — mantisalt · 2026-08-27
- Hermes LCM: Compresses 173k Tokens to 2.8k with Full Retrievability — iamrobotbear · 2026-08-27
- Google releases Gemini 3.5 Transcribe: filters filler words and understands codebase context — AI_Andrew · 2026-08-27
- Why AI agents need stablecoins: traditional settlement takes T+2 to T+60 days — kleffew94 · 2026-08-27
- Migrated docs to Confluence because Notion Cloudflare blocks AI agents — corbtt · 2026-08-27
- Benchmark: OpenCode degrades open model performance, Pi framework leads — PMinervini · 2026-08-27