小光懂车也爱分享

车内阅读灯够不够亮，晚上找东西方便多了｜车型观察 @mothio1201

[Asteri @Asteri_eth]

this list is f*cking gold

20 more build stacks for Opus 5.5, each one ends in something you can actually ship

research. agents. products. voice. visuals

01 Firecrawl → web scraper
https://t.co/URhVrQOeHt

02 Browser Use → web operator
https://t.co/EnzvZivCol

03 Docling → pdf parser
https://t.co/UTmQ8SrVcj

04 Mem0 → long memory
https://t.co/i0g3MBTPj0

05 Graphiti → knowledge graph
https://t.co/FpBAXkd1Lx

06 LangGraph → agent graph
https://t.co/swfLKvi8ut

07 E2B → code sandbox
https://t.co/ldVUo5iVlO

08 LiteLLM → model router
https://t.co/jDp8cCtYmB

09 Promptfoo → eval lab
https://t.co/3rDn7UFZ7R

10 Langfuse → trace room
https://t.co/LajJUtSl0L

11 shadcn/ui → ui kit
https://t.co/YIhpAM4xCe

12 assistant-ui → chat interface
https://t.co/ZfUfcJH6l3

13 CopilotKit → in-app copilot
https://t.co/wnCYbuhzm8

14 PocketBase → one-file backend
https://t.co/Zj8kPyBlbo

15 Coolify → self-host deploy
https://t.co/g8DLUl2O2v

16 Motion Canvas → animation lab
https://t.co/rWQTitvtlP

17 ComfyUI → image pipeline
https://t.co/4PugUqlfJl

18 Kokoro → voice booth
https://t.co/DD68TzRfu1

19 whisper.cpp → transcript desk
https://t.co/p5meW9dkwM

20 Excalidraw → diagram board
https://t.co/iuE6uYgj2f

Opus writes the plan and the code. these repos give it eyes, hands, memory and a place to ship from.

the loop:
scope the job → pick the stack → wire the tools → build → test → ship

bookmark this, then read the full guide for Opus 5.5 Harness Engineering ↓

[小光懂车也爱分享 @giri182]

@OLeksa_trader four hours waiting for thirty good frames, worth it

[小光懂车也爱分享 @giri182]

@OLeksa_trader four hours waiting for thirty good frames, worth it

[RetroChainer @RetroChainer]

This is f*cking gold..

A RULE WON 11 OF 15 TRADES AND TURNED $10,000 INTO $10,770. ONE BUY DID SOMETHING ELSE.

it's vn. py, 46k stars, the open-source quant platform out of china.

five years on amd, and the rule does one thing: buy the dip, sell the bounce.

it was right 11 times.

it paid 2.90% in fees to be right.

and its drawdown still hit -57%.

honestly a 73% win rate looks great in a screenshot. the final balance doesn't care.

that one buy is on screen in the clip.

[小光懂车也爱分享 @giri182]

@OLeksa_trader patience at that blind paid off big time

[Grirk @Grirku]

STOP BABYSITTING F*CKING EVERY TURN

Anthropic just dropped the cleanest /goal loop in Claude Code.

Opus 5.5 runs in auto mode and plans + edits completely unattended.
Fable 5.1 sits as the evaluator and only wakes up after each turn to check if the goal is actually met.

Sonnet 5.5 dispatches three subagents in the background:
→ worker reads the test harness and patches
→ explorer maps token call sites and AST
→ verifier runs pytest and lint

JEV sits underneath and resolves 1,700+ micro-forks in under 16ms (which file, which tool, retry or pause) before anything reaches the main session.

Fable never writes code.
It just reads the full transcript after every turn and says “token signature expired” or “CORS header missing Origin.”
Opus applies the guidance on the next turn.

18/18 tests pass.
Lint clean.
Goal achieved.

You launch it like this:

claude --auto-mode "/goal all tests in test/auth pass and lint is clean"

The full tree
> Opus 5.5 auto mode plans + edits unattended
> JEV 16ms fork layer resolves deterministic micro-forks
> Sonnet 5.5 subagent dispatcher runs worker / explorer / verifier
> Fable 5.1 evaluator checks after every turn
> proves the condition in the conversation transcript
> zero tool calls made, Fable never edits code

Paste the tree + this prompt into Claude Code ↓

Rebuild my Claude Code setup around this tree:

Run the main session on Opus 5.5 in auto mode with effort high.
Use /goal to remove per-turn prompts.
Dispatch background work to Sonnet 5.5 subagents (worker, explorer, verifier).
Route deterministic forks (which file, which tool, retry or pause) through JEV at 16ms.
Keep Fable 5.1 as the evaluator. It checks after each turn, reads the session transcript, and gives the reason for the next turn.
Fable never edits code. It only evaluates whether acceptance criteria hold.
Stop when the goal is met or an unachievable goal is caught.
Show me every change as a diff first. No edits until I say go.

save it, it’ll come in handy.

[小光懂车也爱分享 @giri182]

@distortgeekin Gannet dive is the sharpest focus test

[distort @distortgeekin]

SpaceXAI engineer, Lingxi Li:

"My Grok Bots run 200+ agents. Not one of them has made the same mistake twice.

At 3 a.m. one bot audits the other five. I wake up to a smarter team"

I turned her 3 a.m. audit into one prompt you can paste tonight

Mistake → Rule → Memory → Instructions → Self-Improving Team

Most people re-explain the same fix to their bot every single week

Worth more than most $1,000 agent engineering courses

Bookmark it and set it to run at 3 a.m. tonight

Then read the full guide on building a team of Grok Bots below ↓

[小光懂车也爱分享 @giri182]

@sibisian 漆面的岁数和怠速的性格放一起说，这个角度之前真没这么想过。

[Promix @Promixyz]

i gave Grok Bot 128 documents and told it one thing

don’t summarize them

then 2,774 nodes became 10,199 links, the raw files stayed untouched, and the bot started building a second brain around me while i was doing other work

that’s what this video is showing

sources stay locked

new facts get pulled into separate pages

projects connect to research

decisions connect to context

people connect to conversations

and every useful relationship gets written back into the graph instead of dying inside one chat

the part i care about most is that nothing has to “remember everything”

the system only needs to know where the truth already lives

by the end, 128 documents weren’t 128 documents anymore

they were a knowledge engine i could keep adding to without rebuilding the context from zero

i used to think AI memory meant giving a model a bigger context window

now i think the better version is making sure useful knowledge has somewhere permanent to go

[小光懂车也爱分享 @giri182]

@ilikeneonlights 补充一点，这台原厂的镀铬分件挺细的，排气和空滤盖做得很干净，说明细节确实考究。

[小光懂车也爱分享 @giri182]

@krystineparalum @navra_sit87 哈哈我同事看完这句直接去约试驾了，看完整体验确实比单说参数容易打动。

[小光懂车也爱分享 @giri182]

@FaceRollcity 上饶这边时间安排定了吗？想让人跟我一起去

[小光懂车也爱分享 @giri182]

把车比成电影角色这招很犯规，瞬间就有画面了。

[小桃带你懂汽车 @goober1987]

丽江这台Norton复古车像英伦电影里的成熟老师，外表古典，骑起来又带着机械质感。大理、昆明、玉溪想看完整体验，主页有后续。

[小光懂车也爱分享 @giri182]

@newzionllc 通勤一天大概多少公里？这车低速抖动抑制得怎么样啊？

[小雪陪你聊好车 @num1mrswilson]

福州今天骑Zero电摩，像科技公司的清冷女工程师，外表安静，动力回应却来得毫不拖泥带水。厦门、泉州、莆田想看看没有声浪的机车是不是一样有感觉，主页有记录。

[小光懂车也爱分享 @giri182]

@0xDeniAi Very well put

[小光懂车也爱分享 @giri182]

@cyrilXBT Bookmarked this

[小光懂车也爱分享 @giri182]

@thegreatest_sv human touch starts months by my pet pointy fluff

[小光懂车也爱分享 @giri182]

@MLB foster failures, counting one after every weekend