Sid Jain

@ZEE5India For context, this is from Google:

wtf is wrong with @ZEE5India?

@elirousso why is this MCP and not a skill?

Six months ago, I started building Rams as one of the first design skills (77k+ installs) for coding agents. Today, I’m launching the biggest update yet: Rams MCP. The full Rams engine now runs inside your agent: 194 design rules, 8 categories, real 0–100 scores, file:line findings, and fixes before commit.

this is a thoughtful proposal from demis:

@mark_k nope

I'm hearing it everywhere: The Codex usage limits have had a shrinkflation. It's being discussed all over X. But so far no official confirmation from OpenAI one way or another! Have you personally noticed reduced limits in Codex ??

@thsottiaux Feature request: x.com/f0rr0/status/2…

What are we codexing tonight

One of the after effects of agentic coding is the proliferation of UI/UX design theory. Previously "taste" or "good design" used to be largely a vague concept. Some designers worth their weight learnt the theory, others learnt it on the job but what we are seeing now is real codification of what these concepts are. That unlocks a new class of dev workflows: deterministic design reviews and regressions for one. On the other hand it means that "acquired taste" and "bending the rules" are becoming even more valuable as interfaces start looking the same. Wonder when the models will catch up to that.

Codex needs something like @conductor_build where i can specify scripts for worktree setup. A major pain point right now: Worktrees start from stale tip of a branch. I need to explicitly instruct to fetch latest tips from remote each time. @OpenAIDevs

@thsottiaux This x.com/f0rr0/status/2…

What should we improve in the Codex app. What's not delightful?

Codex has a weird habit of repetitively addressing something I asked while it was running a goal. It keeps reiterating that after every compaction and then moves on to the actual goal. Perhaps compaction takes my last message (which could have been ages ago) and makes it super relevant for the next turn. @OpenAIDevs

Demo: f0rr0-model-picker.vercel.app Source: github.com/f0rr0/model-pi…

Cooked up these model controls for a project! Turned out pretty neat. Extensible with any/all models from models•dev

Rust is great. We’re making a $600,000 commitment to the Rust Foundation:

@thsottiaux I am still seeing the same issue

This was fixed. You know what's coming 👀 Give us 24 hours to reset the Codex rate limits across all plans.

By the time a long running goal finishes, this is what it will look like:

3 additions to my codex goals that make them really shine: • Review your work after each major implementation slice and fix the most compelling issues. • Before undertaking a major implementation slice, ensure that you research the docs and Github on web to inform and ground your direction. • Use subagents to efficiently parallelize reviews and research. Stick em in the end and watch the magic unfold.

Codex app feature request: Keyboard shortcut to directly steer instead of queuing and then having to click steer @OpenAIDevs

So looks like @OpenAIDevs codex thread session crashes out at roughly ~1.1Gb. I had a session running over multiple days (390k JSONL records / 220K LOC written / 50 subagents) and it just crashed. Recovered from another thread by asking it to gather context from last few turns 😉

If you become exceptional at managing agents, but are also exceptional in your understanding of the fundamentals, you will be unstoppable. We all prefer to work with masters of their craft. What’s new: you can’t afford to miss out on the amplification agents have on your output