Sabitlenmiş Tweet
Christopher Man
1.4K posts
Christopher Man
@int32max
chris - senior product engineer - @retailbook_ - https://t.co/II5CSYCHtH
London Katılım Nisan 2016
929 Takip Edilen243 Takipçiler
Finally don't have to look at tiny text in termux with ssh
OpenAI Developers@OpenAIDevs
Step away from your laptop. Keep building with Codex on your phone. Codex keeps working on your computer, with your files and project context still in place. Pocket-sized access. Full Codex working state. x.com/OpenAI/status/…
English
@TheJuanMaggi @justalexoki tailscale with tmux is so easy to setup, just point codex at your homelab or mac and ask it to set it up
agree with you here
English
@justalexoki Tailscale is all you need bro, just ssh into your computer and use codex just fine. I literally control codex, gemini, opencode all through my phone whenever I'm away from the house. Super easy to setup.
English
@iyoushetwt 4 years straight I ran arch with i3
some of my best times spent on a machine
soooo much time doing configs, never did anywork
being employed out of uni forced me to use a macbook to get shit done
linux for homelab is the only sane choice
English
the ultimate combo
even better one is walk down the thames with a coffee
Numman Ali@nummanali
@int32max Vape and Coffee break
English
looking at my ssh session into codex while im vaping in the parking lot
Beff (e/acc)@beffjezos
when you're jetlagged af but your homie sends you a niche banger
English
@joshmanders exactly proper ide with agent on the side will always win
English
@yacineMTB still confused as to why we need robots to look humanoid like
i'd be happy with a big metal box that did sorting (would probably do it more efficiently aswell)
English
this doesnt include my 2-3 hours a night working on my game engine
spend the commute reading books on my jail broken kindle
GIF
English
double my effective time at work with strict work blocks and strict break times
solid breaks means more cognitive refresh times so I can stay more focused when working with coding agents
been working more then I ever have without working 'more' just significantly more efficient
never used to take breaks but having proper dedicated break times actually makes me more effective
simple time blocks is better then anything i've done before (pomodoro doesnt work, time iterations too short to actually get work done)
simplicity scales complexity doesnt
brain checks calendar -> need to work or break ? -> execute work or break, easy as fuck
English
multiple day session of codex running gpt 5.5 high and its been perfect with auto compaction
codex to use subagents for reviews and /side is great for asking explanation and clarifying questions
with claude I was always worrying about context window after 350-400k context window
never think about it now with codex cli locally
English
yeah you cant tell me a full HTML page to explain this is any more effective then inline uml text generation
less context and easier to parse imo
Christopher Man@int32max
as a extremely visual based learner i am constantly getting the agent to spin up local UML diagrams for explainations attempted the html route but found it took too long and the extra 'polish' it had was actually more confusing gimme dirty simple flow/UML diagrams in chat everytime when it does get bigger I end up using mermaid diagrams in markdown for things that need saving
English
we are soooo back
switched to codex fully couple weeks ago at work and on personal projects
never looked back
excited to see the turnout today
English
@verrsane working in london now
when i compare the salaries to some of my friends in SF its insane the salary difference
GIF
English
My wife used to work at a UK bank and said her coworkers used to flag accounts with a couple thousands in them….
These people are broke asf😭
Markov@MarkovMagnifico
holy shit there’s no way the UK is actually this poor
English
"itchigo found 2 CVE's in the codebase"
"madara is working on a refactor of auth"
GIF
English
i'm so easily amused by the names that codex give its subagents
wonder if its configurable so I can give them names of anime characters
English
as a extremely visual based learner i am constantly getting the agent to spin up local UML diagrams for explainations
attempted the html route but found it took too long and the extra 'polish' it had was actually more confusing
gimme dirty simple flow/UML diagrams in chat everytime
when it does get bigger I end up using mermaid diagrams in markdown for things that need saving
English
Can I just be for real for a sec
Vscode/zed + cc/codex-cli/pi
Gives you entire ide with everything you need.
As someone who was obsessed with custom harness I just don't care anymore
I need a feature full ide to get work done
Lowkey it's like my arch i3 -> macos journey
Reliable tools every time.
Also benefit of doing this is that you get the harness for the agent maintained by the model provider which means sub subsidized costs
for now...
ClaudeDevs@ClaudeDevs
Starting June 15, paid Claude plans can claim a dedicated monthly credit for programmatic usage. The credit covers usage of: - Claude Agent SDK - claude -p - Claude Code GitHub Actions - Third-party apps built on the Agent SDK
English
tanstack query is so nice to use, can fully offload server/api state
it just works without headache, another piece of beautiful software
English
one builder and one explainer agent
hard to build fast without having the knowledge of what the agents are actually doing
need to build this greenfield project out for stakeholders in the next 4 weeks
shipping speed is great but knowledge of how its working is just as important to keep up as the project grows
English