This daily picks page was compiled by the Postprint pipeline on 2026-10-05 (Asia/Shanghai). Sampling covers 20 public X accounts (including @pmarca, @balajis, @karpathy, @paulg, @sama, @naval and others) through the public read-only FxTwitter API — this is an account-pool observation, not an X-wide ranking. Today's sampling observed 95 items; 4 of them were newly verified and published on this site today; 1 were excluded by our editorial content filter. Entries below are ordered by a composite pick score (likes + 2×reposts + 3×bookmarks + 0.001×views) and presented as editorial selections. All interaction numbers are real sampled values shown with their collection time (2026-10-05 UTC); metrics we could not collect are omitted, never filled with zeros.
On this page
- 1. We'll be spending a lot more time trying to understand the outputs of language models.
- 2. We're publishing our most detailed threat intelligence report to date.
- 3. One pattern I find useful for working with LLMs is a nice long ramble session.
- 4. This works really well btw, at the end of your query ask your LLM to "structure your response as HTML", then…
- 5. A harness for every task: dynamic workflows in Claude Code
- 6. This is a new paradigm for interacting with Claude that is significantly more "inline" with all the other…
- 7. The AI Agent Complexity Ratchet: Why 90% Test Coverage Is Required
- 8. YC as a Witness
- 9. Own your AI, don’t let it own you
- 10. Here's my conversation with Don Lincoln about some of the biggest open questions in physics, including dark…
1. We'll be spending a lot more time trying to understand the outputs of language models.
Andrej Karpathy (@karpathy) · post · sampled 2026-10-05
52,461 likes6,102 reposts1,495 replies77,491 bookmarks1,330 quotes6,974,123 views
metrics sampled 2026-10-05 (UTC)
What it is: A long X post published on X by Andrej Karpathy (@karpathy), about 1,816 characters of text. It opens with: "We'll be spending a lot more time trying to understand the outputs of language models. A few thoughts, tips & tricks: Writing. Something I've had success with: Ask your LLM to explain something in ASD-STE100, it's a controlled language specification originally developed for…" The headline above is our summary of the post's opening, not the author's own title.
Why it is getting attention: From these sampled numbers, the sampled view count (6,974,123) suggests it reached a wide audience within this account pool; the sampled like count (52,461) is well above the rest of today's sample. This is an editorial observation based on publicly visible metrics, not an official ranking or platform conclusion.
Original on X ↗ Export it with Postprint →
2. We're publishing our most detailed threat intelligence report to date.
Anthropic (@AnthropicAI) · post · sampled 2026-10-05
50,530 likes11,711 reposts3,231 replies41,932 bookmarks4,931 quotes43,732,266 views
metrics sampled 2026-10-05 (UTC)
What it is: A long X post published on X by Anthropic (@AnthropicAI), about 936 characters of text. It opens with: "We're publishing our most detailed threat intelligence report to date. It covers how people tried to misuse Claude—for cyberattacks, influence operations, surveillance, biology, and building weapons—and how we found and stopped them. We disrupted every operation in the…" The headline above is our summary of the post's opening, not the author's own title.
Why it is getting attention: From these sampled numbers, the sampled view count (43,732,266) suggests it reached a wide audience within this account pool; the sampled like count (50,530) is well above the rest of today's sample. This is an editorial observation based on publicly visible metrics, not an official ranking or platform conclusion.
Original on X ↗ Export it with Postprint →
3. One pattern I find useful for working with LLMs is a nice long ramble session.
Andrej Karpathy (@karpathy) · post · sampled 2026-10-05
50,125 likes4,081 reposts2,521 replies15,119 bookmarks1,490 quotes4,362,430 views
metrics sampled 2026-10-05 (UTC)
What it is: A long X post published on X by Andrej Karpathy (@karpathy), about 797 characters of text. It opens with: "One pattern I find useful for working with LLMs is a nice long ramble session. Sometimes the LLM needs more bits to understand what you're trying to achieve, but you're too lazy to type them. In these cases I like to lean back, switch to /voice and just ramble for like 10…" The headline above is our summary of the post's opening, not the author's own title.
Why it is getting attention: From these sampled numbers, the sampled view count (4,362,430) suggests it reached a wide audience within this account pool; the sampled like count (50,125) is well above the rest of today's sample. This is an editorial observation based on publicly visible metrics, not an official ranking or platform conclusion.
Original on X ↗ Export it with Postprint →
4. This works really well btw, at the end of your query ask your LLM to "structure your response as HTML", then…
Andrej Karpathy (@karpathy) · post · sampled 2026-10-05
19,432 likes2,030 reposts1,053 replies21,480 bookmarks484 quotes4,069,759 views
metrics sampled 2026-10-05 (UTC)
What it is: A long X post published on X by Andrej Karpathy (@karpathy), about 1,916 characters of text. It opens with: "This works really well btw, at the end of your query ask your LLM to "structure your response as HTML", then view the generated file in your browser. I've also had some success asking the LLM to present its output as slideshows, etc. More generally, imo audio is the…" The headline above is our summary of the post's opening, not the author's own title.
Why it is getting attention: From these sampled numbers, the sampled view count (4,069,759) suggests it reached a wide audience within this account pool; the sampled like count (19,432) is well above the rest of today's sample. This is an editorial observation based on publicly visible metrics, not an official ranking or platform conclusion.
Original on X ↗ Export it with Postprint →
5. A harness for every task: dynamic workflows in Claude Code
Thariq (@trq212) · article · sampled 2026-10-05
11,042 likes1,464 reposts276 replies24,785 bookmarks307 quotes3,392,594 views
metrics sampled 2026-10-05 (UTC)
What it is: A long-form X article published on X by Thariq (@trq212). It opens with: "Last week, we released dynamic workflows in Claude Code. Claude can now write its own harness on the fly, custom-built for the task at hand. While the default Claude Code harness is built for coding,"
Why it is getting attention: From these sampled numbers, the sampled view count (3,392,594) suggests it reached a wide audience within this account pool; the sampled like count (11,042) is well above the rest of today's sample. This is an editorial observation based on publicly visible metrics, not an official ranking or platform conclusion.
Original on X ↗ Export it with Postprint →
6. This is a new paradigm for interacting with Claude that is significantly more "inline" with all the other…
Andrej Karpathy (@karpathy) · post · sampled 2026-10-05
23,514 likes1,945 reposts1,383 replies13,939 bookmarks706 quotes8,400,391 views
metrics sampled 2026-10-05 (UTC)
What it is: A long X post published on X by Andrej Karpathy (@karpathy), about 853 characters of text. It opens with: "This is a new paradigm for interacting with Claude that is significantly more "inline" with all the other human activity org-wide. Once you do all of the under the hood engineering work to make this "just work" (e.g. across tools, integrations, compute environments, memory,…" The headline above is our summary of the post's opening, not the author's own title.
Why it is getting attention: From these sampled numbers, the sampled view count (8,400,391) suggests it reached a wide audience within this account pool; the sampled like count (23,514) is well above the rest of today's sample. This is an editorial observation based on publicly visible metrics, not an official ranking or platform conclusion.
Original on X ↗ Export it with Postprint →
7. The AI Agent Complexity Ratchet: Why 90% Test Coverage Is Required
Garry Tan (@garrytan) · article · sampled 2026-10-05
793 likes95 reposts50 replies1,571 bookmarks37 quotes244,071 views
metrics sampled 2026-10-05 (UTC)
What it is: A long-form X article published on X by Garry Tan (@garrytan), about 20,419 characters of text, 1 image(s). It opens with: "I've been coding with AI for the past year. Not just prompting -- building real software. Two open-source projects: GStack, which makes AI coding agents better, and GBrain, which turns everything you read and write into a searchable knowledge base your AI can use. Between…"
Why it is getting attention: From these sampled numbers, the sampled view count (244,071) suggests it reached a wide audience within this account pool; the sampled like count (793) is well above the rest of today's sample. This is an editorial observation based on publicly visible metrics, not an official ranking or platform conclusion.
Read & export this article on Postprint →
8. YC as a Witness
Garry Tan (@garrytan) · article · sampled 2026-10-05
1,059 likes137 reposts60 replies1,358 bookmarks81 quotes486,934 views
metrics sampled 2026-10-05 (UTC)
What it is: A long-form X article published on X by Garry Tan (@garrytan), about 11,175 characters of text, 1 image(s). It opens with: "The psychologist Alice Miller spent her career studying one question: why do some people who grow up in brutal circumstances become whole adults while others repeat the cycle?"
Why it is getting attention: From these sampled numbers, the sampled view count (486,934) suggests it reached a wide audience within this account pool; the sampled like count (1,059) is well above the rest of today's sample. This is an editorial observation based on publicly visible metrics, not an official ranking or platform conclusion.
Read & export this article on Postprint →
9. Own your AI, don’t let it own you
Sigil Wen (@0xSigil) · article · sampled 2026-10-05
1,085 likes123 reposts93 replies1,253 bookmarks42 quotes443,209 views
metrics sampled 2026-10-05 (UTC)
What it is: A long-form X article published on X by Sigil Wen (@0xSigil), about 7,581 characters of text, 1 image(s). It opens with: "6 months ago, the #1 AI was Claude Opus 4.6. Our model, Underdog 27B beats it while running on your Mac or iPhone. Most AI will run on your device, not in data centers."
Why it is getting attention: From these sampled numbers, the sampled view count (443,209) suggests it reached a wide audience within this account pool; the sampled like count (1,085) is well above the rest of today's sample. This is an editorial observation based on publicly visible metrics, not an official ranking or platform conclusion.
Read & export this article on Postprint →
10. Here's my conversation with Don Lincoln about some of the biggest open questions in physics, including dark…
Lex Fridman (@lexfridman) · post · sampled 2026-10-05
2,091 likes277 reposts207 replies833 bookmarks36 quotes265,035 views
metrics sampled 2026-10-05 (UTC)
What it is: A long X post published on X by Lex Fridman (@lexfridman), about 1,477 characters of text. It opens with: "Here's my conversation with Don Lincoln about some of the biggest open questions in physics, including dark energy, dark matter, the matter-antimatter imbalance, quantum vacuum, quantum foam, and the quest to unify the laws of physics."
Why it is getting attention: From these sampled numbers, the sampled view count (265,035) suggests it reached a wide audience within this account pool; the sampled like count (2,091) is well above the rest of today's sample. This is an editorial observation based on publicly visible metrics, not an official ranking or platform conclusion.
Read & export this article on Postprint →
The takeaway
Taken together, today's batch is 10 long-form pieces drawn from the same public-account pool, mostly articles and threads saved and shared by readers. The full, continuously updated collection lives on the trending page, and any X article or thread can be turned into a document with the Postprint tool.
Source coverage: 20 public X accounts sampled via FxTwitter (public, read-only). This page reflects only what those accounts published and what our sampler could verify — it is not affiliated with, endorsed by, or an official ranking of X. Sampling date: 2026-10-05 (Asia/Shanghai). Questions or takedown requests: contact us.
Snapshot note: this page preserves the metrics as sampled on
2026-10-05; later re-runs refresh the same URL for the same day and never rewrite
past days' snapshots.