IngoA (@ingoa_dev) header
IngoA (@ingoa_dev) profile picture

IngoA

@ingoa_dev

I'm doing the thing w mobile apps.

 Yourope
Joined November 2022
  • Tweets14.3K
  • Following1.2K
  • Followers2.4K
638Media
Photo by IngoA (@ingoa_dev)Photo by IngoA (@ingoa_dev)Photo by IngoA (@ingoa_dev)Photo by IngoA (@ingoa_dev)Photo by IngoA (@ingoa_dev)Photo by IngoA (@ingoa_dev)Photo by IngoA (@ingoa_dev)Photo by IngoA (@ingoa_dev)Photo by IngoA (@ingoa_dev)Photo by IngoA (@ingoa_dev)
IngoA retweeted
ani (@anaisbetts) profile picture
ani@anaisbettsNov 4
quick question, what the fuck
ClaudeDevs (@ClaudeDevs) profile pictureClaudeDevs
@ClaudeDevsNov 4
Background subagents in Claude Code now use less of your usage limit. A running subagent updates its status line every 30 seconds, and each update used to re-read its whole conversation. Now Haiku 5.5 writes it from short notes instead, so each update costs about 99% less.
51
74
4
5,256
474,643
Supposedly this does not work anymore since last Codex update. I overheard a prompt that may fix it, in the replies.
🥔🥔🥔 (@argofowl) profile picture🥔🥔🥔
@argofowlOct 3
everyone thinks codex computer use only works inside codex it doesn't, it's a local mcp server in the chatgpt mac app and claude code can just call it, even headless with claude -p same 8-task test: opus 5.5 on codex's engine got 6/8, codex itself got 6/8, cua driver got 3-4/8 at 4x the cost cua's background mode on mac only goes through accessibility, so canvases and drags just don't land codex's engine sends real clicks and drags to the app without touching my cursor no hover though, and it's unofficial, so enjoy it until a chatgpt update breaks it give this to your claude: "Set up Codex's computer use as an MCP server for you on my Mac. Find the "cua_repl" entry in ~/.codex/plugins/cache/openai-bundled/unified-computer-use/<newest version>/.mcp.json and register it as a user MCP server called "codex-cu" with the same command, args and env. Then test it by using Calculator in the background to work out 12 × 12."
2
2
266
more replies
Rename the helper .app so Spotlight can tell it apart. Copy the plugin's .mcp.json and repoint every ChatGPT.app and ~/.codex path, including require-browser-access.mjs. Set CODEX_HOME, CODEX_CLI_PATH, PATH, SKY_CUA_SERVICE_PATH and SKY_CUA_SERVICE_NATIVE_PIPE_PATH.
1
32
Run the helper from a LaunchAgent with those same vars. It needs codex itself, or you get -10005 "app-server exited". The private socket lets it run beside ChatGPT's helper. Test by driving an app, not listing tools: ChatGPT closed, ChatGPT open, then after a reboot.
20
The Aliens edit each other now.
Rikiya Takehi (@rikiyatakehi) profile pictureRikiya Takehi@rikiyatakehiNov 2
We asked GPT-6 Astra to edit Qwen3.5-0.8B weights to improve math. It directly changed only 0.016% of weights, fixing 0.8B's reasoning loops & improving math up to 6x and even instruction following 3x, without gradients or training. We call this Weight-Space Programming: frontier LLMs can now natively read/write neural network weights! 🧵 Blog: rikiyat.github.io/blog/weigh…
Image from Rikiya Takehi (@rikiyatakehi)
Readers added context they thought people might want to know
LLM was given weights explicitly as text in prompt and used code/forward passes (per authors' logs) to ablate Qwen looping direction, a known thinking-mode issue per model card. Primarily fixed repetition in that mode, not general 'natively' improving weights/math. rikiyat.github.io/blog/weight-sp… github.com/RikiyaT/lm_und… huggingface.co/Qwen/Qwen3.5-0… huggingface.co/RikiyaT/Qwen3.…
95
I'm talking to my therapist right now. Fun app, japan only 🤷‍♂️
Anum  (@anumness) profile pictureAnum 
@anumnessOct 20
I really hate that I cant build new apps. Building apps was therapy for me. Im having an existential crisis.
1
195
Opus being lazy and dumber for me since yesterday, anyone else, did the nerf begin?
2
4
240
Hopefully it's just the Fable 5.5 release coming up.
1
98
Slop everywhere! Prevent weird conflicts with AGENTS .md, and stay DRY: "autoMemoryEnabled": false Claude no longer reads or writes the memory folder. "autoDreamEnabled": false the background step that tidies memories is off as well. in ~/.claude/settings.json
3
190
Cursor for video is here, BTW. Name is Opus 5.5
140
Claude Desktop's task list and wrap-up boxes confusing and in your way? Add to ~/.claude/CLAUDE.md: "Don't use the task list (update_plan) or the wrap-up box (wrap_up). Reply in plain linear chat; ask questions in the message text." Probably keeps context cleaner too 🙌
194
"A few vblanks" - good old times!
John Carmack (@ID_AA_Carmack) profile pictureJohn Carmack
@ID_AA_CarmackOct 16
I had a similar reaction. Everyone should applaud performance improvements, but gauge against the “speed of light”, not just a sloppy competitor. If you are measuring in seconds, you may have already anchored yourself incorrectly. Structure the UI so you can put locally cached media up just a few vblanks after the process launches. Structure the server so it can return everything necessary for the dynamic UI in one single query that was kicked off in parallel with the Gfx API initialization, branching to a modern transport protocol for large media files. You should be able to double click on the launcher and have it responsive before your muscle memory can slam the pointer to a location to click on.
1
101
End of an era 🥲 (500 requests plan since Sept 2023, GPT-4 times)
Image from IngoA (@ingoa_dev)
1
1
123
There's no turning back to this plan. Good times going for 2M token mega-prompts with Opus 4.6 :)
30
. @steipete included the follow-claude-desktop feature into CodexBar, as per my tickets ❤️
IngoA (@ingoa_dev) profile pictureIngoA@ingoa_devOct 14
I made a tiny macOS usage bar for Claude Desktop. It follows whichever account you're signed into, all accounts at a glance. Couldn't find one that did this, so I built it: github.com/flutterrausch/cla…
Image from IngoA (@ingoa_dev)
4
335
Not a single use case in the comments.
Jon Yongfook (@yongfook) profile pictureJon Yongfook
@yongfookOct 15
People who are running "agent swarms in the cloud" while they sleep, what's the use case? Even if I had a backlog of 100 issues I'm pretty sure a single Claude instance could rip through that in 30 mins let alone multiple, all night. Curious to know what people use this for.
2
186
my life before sama's tweet:
Sam Altman (@sama) profile pictureSam Altman
@sama30 Nov 2022
today we launched ChatGPT. try talking with it here: chat.openai.com
40
185
9
2,873
104,726
You better turn off this "lose context" thing.
Lydia Hallie ✨ (@lydiahallie) profile pictureLydia Hallie ✨
@lydiahallieOct 14
We recently added idle compaction to Claude Code! It now compacts long idle sessions right before the prompt cache expires (~1h). If you come back to it later, it only has to reprocess the summary instead of the entire conversation, which saves you some usage.
Image from Lydia Hallie ✨ (@lydiahallie)
97
I made a tiny macOS usage bar for Claude Desktop. It follows whichever account you're signed into, all accounts at a glance. Couldn't find one that did this, so I built it: github.com/flutterrausch/cla…
Image from IngoA (@ingoa_dev)
3
1
3
602
Most usage trackers are built for Claude Code, not the desktop app. They don't follow its account switches, so they keep showing the old account. Greetings to your clanker.
1
54
Most of the effort went into legal slop. Naming (Clawd ...), forbidden features..
27
Load more