Lance Herron

opencode 1.3.0 will no longer autoload the claude max plugin we did our best to convince anthropic to support developer choice but they sent lawyers it's your right to access services however you wish but it is also their right to block whoever they want we can't maintain an official plugin so it's been removed from github and marked deprecated on npm appreciate our partners at openai, github and gitlab who are going the other direction and supporting developer freedom

@bentossell Sell side makes perfect sense. Now explain buy side.

moltbook aquisition if you vibed something in days and could sell it to meta for a lot of money you would fair play for getting the bag. this games all about opportunities and agency

@HamelHusain Many such cases. So many….

The good/bad part about agentic codeing is the barrier to getting nerdsniped is now much lower github.com/hamelsmu/hamel…

@deanwball @benthompson @FifthRocket @Noahpinion Once you two have your positions solidified, please do provide some public discourse. It’s a nuanced topic and I see a lot of “Ben should stay in his lane” takes in other threads. This seems like the right way to air it all out.

@benthompson @FifthRocket @Noahpinion Done.

Honestly, in the @benthompson vs @deanwball debate, I think Ben is right. There was just no way America -- or any nation-state -- was ever going to let private companies remain in total control of the most powerful weapon ever invented.

@tunguz Need more side-projects, obviously.

idk, i feel that all this agentic coding is exacerbating my doomscrolling addiction. i mean, i can’t do anything else that’s meaningful between waiting for the coding agents to finish their 15-20 minutes bursts of work.

@andrew_n_carr Everything is Claudeputer.

I didn't write any code this weekend, nor did I have agents write any code. I took space to think, write notes, walked a bunch. It was honestly the worst. Need to be Claude maxxing...

@voooooogel Likely unrelated to your issue but I’ve definitely noticed some 4.6 degradation ever since the harness change (and I assume system prompt update) to use Explore subagent on practically every user prompt.

has anyone else switched their default back to opus 4.5 in claude code? i'm considering it. 4.6 is good at some things but i keep running into situations where it anxiously rabbitholes / thrashes on problems that 4.5 would've handled fine. could also be a harness change?

@Presidentlin I plan with 5.2-pro and then play RTS. Feels like the right balance for me personally.

I'm thinking of going back to Stage 2/2.5 of Steve Yegge's 8 Stages of Dev Evolution to AI I've lost control of my code base. I miss being able to dream about the code, but I can't go back to 2022 slow style coding. Yegge Stage 2 was Coding agent in IDE, mine is coding with ChatGPT Projects and copy-paste code manual. Stage 2.5 is using Projects for the plan mode and taking that spec to the CLI and churn out the work. Stage 3 for me is you ditch the Projects flow and just use the CLI agent directly, heavy use of plan mode/steering, etc. After Stage 4, you can kiss code control goodbye. There are a few exceptions for Front end, I'll one-shot with Gemini CLI and frontend skill, but the copy will be done in Stage 2. Stage 2 is perfect since it's the Copilot era. Stage 2 is the closest I'll get to the 2022 world. My stages are: 1 - AI in Editor 2 - Coding with ChatGPT, DeepSeek, or some ChatApp 3 - Coding with 1 CLI Agent in Code Editor 4 - Coding with two Code Editors VSCode & Antigravity 5 - Crazy setup with many CLI/Ralph Loops/RTS Game 6 - Switching from CLI to Agent Manager 7 - Probably some Slack/WhatsApp-based thing connected to stateful sandboxes, Kanban fits. 8 - Not invented yet

The big labs should report stats on user-generated prompt sizes. The more I use gpt-5.2-pro, the more I find myself writing larger and larger prompts to steer the work. May just be the async nature of dealing with pro response times.

@badlogicgames I made this too! But I set it up with a “target process” so I could pipe transcribed keys direct to a target tmux session (running Pi/CC) when I have another window focused.

@theLance we just used my machine. i have my own whisper based transcription app. press button, talk, release button, transcription gets inserted in active window.

today i taught a PM how to use a coding agent. left him alone after 30 minutes, only him + my voice transcription app + pi. he build a standalone version of a feature his old team needed 2 months for, because the struggled reversing an API. the agent figured it out in 10 minutes. then build him a frontend plus netlify function and deployed it. i'm actually not making this up or exagerating. is it production ready? of course not. but that doesn't matter. as a consultant, he can now build functional prototypes himself together with customers. once they have figured out how things should work, they can hand it off to a software firm to build it properly. possibly with agents as well, thereby (hopefully) reducing time and cost. probably hard to quantify such unlocks on a macro level, but this should contribute meaningfully to growth imo.

@badlogicgames Did you set up voice/pi for them? I think often the hurdle is getting the environment ready for them as much as teaching them how to use it.

this is a repeating pattern i've seen since may 2025. trach an expert the basics of a coding agent, taking into account their domain, and they can get productive really fast. what i haven't figured out yet is how to scale up that teaching effort. my guess is, that an agent could teach the user. but as i said yesterday, it's super hard to build learning paths with agents.

@badlogicgames Very cool!

🤗 π

@sebkrier It’s the best timeline.🍿

It seems like we're due another huge wave of very original tweets about AI progress continuing, read primarily by people who already know that AI progress is continuing! 😏

Only complaint so far - something is really off on the scrolling logic. I'm constantly having to scroll down because the agent turns are scrolling off the bottom of the screen.

Really enjoying the Codex app! It's a bit of a mind shift from "new terminal window for every session" to "new chat thread for every session" but it makes spinning up parallel work feel much more natural.

@iruletheworldmo Maybe we get skynet because megacorp tech ceos can’t get along.

kinda worrying watching this given these two are building asi. i’m hoping the asi would take the hand.

@dexhorthy @trq212 I hope you're right and they clarify that 3 is fine in writing in the ToS. But if they are supply constrained and they know they can sell all those tokens with full margins via API, seems more likely that they will hobble the subscription-based plans as much as possible.

Clarification and My read: I am not a lawyer, based on @trq212 clarifications today 1 don't use max plan in other coding agents 2 don't use the max plan in openclaw 3. If you use the max plan for personal projects and you use the Claude agent sdk that's ok 4. If you are building a saas, do not serve other people inference using a max plan auth token, even if via the sdk So I think usage of codelayer or riptide or conductor etc is closest to 3 but it’s technically not well defined. However as @0xblacklight said we are preparing many alternatives in case this becomes an issue including finding cheap inference and passing through cost, allowing open code / codex sdks etc If it turns out 3 is fine and anthropic is mostly worried about 4, then yes actually I think they are in their fair rights to try to put a stop to 4 and I wouldn't hate on them for it publicly

@JohnnyNel_ Me too!

@theLance building in public with AI tools forced me to document better than any security protocol ever could

Up next, no Claude Code inside tmux. You can only run CC on bare metal using a wired terminal with a camera recording the human as they type.

@thsottiaux People would love to have hooks, we've been asking ever since the CLI released.

Codex team is fairly distributed, but most of the team is gathering in person over next 48 hours to take a step back and align on what’s next this year. What should we discuss?