Mark Witt

@Dimillian @gabrielchua why specifically green?

Thanks to @gabrielchua, the latest version of the image gen skill will know how to generate spritesheet with green backgrounds so Codex can then have sprites with transparent background and integrate them into your games!

@jmeistrich @zeddotdev are you using the Agent or ACP?

Cancelled my Cursor subscription. I've been testing Zed for the past week and quite liking it! Big props to @zeddotdev, it's come a long way!

@gregisenberg Nope. We‘re building the opposite

It feels pretty obvious at this point that someone’s going to make billions building a social app that’s just for friends, no AI slop, no brainrot, calm design, chronological feed and no concept of followers

@MicrosoftAI How does it compare on benchmarks vs other models?

MAI-Voice-1 exports flawless audio in record speed. Can you tell the difference between the real recording and our model? Drop your best guess. Try it today. msft.it/6018QfcD8

@OfficialLoganK Does it support native timestamping?

Introducing Gemini 3.1 Flash TTS 🗣️, our latest text to speech model with scene direction, speaker level specificity, audio tags, more natural + expressive voices, and support for 70 different languages. Available via our new audio playground in AI Studio and in the Gemini API!

@wesbos when will this release realistically?

I don't think we're ready for an explosion of in-browser text effects like this

@MistralDevs why do you only provide voices in 3 languages although you support 9 Languages?

🎙️Choose your ideal voice and bring your text to life.

@bcherny is there any advantage of doing this as CLI? why reinvent the wheel?

Today we're excited to announce NO_FLICKER mode for Claude Code in the terminal It uses an experimental new renderer that we're excited about. The renderer is early and has tradeoffs, but already we've found that most internal users prefer it over the old renderer. It also supports mouse events (yes, in a terminal). Try it: CLAUDE_CODE_NO_FLICKER=1 claude

@ExaAILabs pricing?

Introducing Exa Monitors - your agent’s radar for the web Exa is a search engine built from scratch, and today we're exposing our "update" layer. Simply define what to find and how often - Monitors will return any new information, via webhook.

@candyflipline in production?

Being early to DSPy is being early to winning. Every agentic logic we have is wrapped in DSPy functions.

@birch_js On iOS its easy developer.apple.com/documentation/…

I keep telling people how powerful the web platform is. Good luck implementing this in React Native

@tejask @expo @ccheever @Baconbrix what are you building? 👀

for real, @expo is the best developer tool and an absolute Godsend coupled with EAS for distribution, the entire mobile + native + web development lifecycle for web-first devs is solved. really really proud of the great work @ccheever, @Baconbrix, and the team have done.

@MistralAI I can only find 2 languages how to enable 9?

🔊Introducing Voxtral TTS: our new frontier open-weight model for natural, expressive, and ultra-fast text-to-speech 🎭Realistic, emotionally expressive speech. 🌍Supports 9 languages and accurately captures diverse dialects. ⚡Very low latency for time-to-first-audio. 🔄Easily adaptable to new voices

@kimmonismus 9 languages? i only see english and french on it

Mistral AI released Voxtral TTS, a 3-billion-parameter text-to-speech model with open weights that the company says outperformed ElevenLabs Flash v2.5 in human preference tests roughly 63% of the time on standard voices and nearly 70% on voice customization. The model runs on about 3 GB of RAM, achieves 90-millisecond time-to-first-audio, supports nine languages, and can clone a voice from just five seconds of reference audio, including cross-lingual adaptation that preserves the speaker's accent.

@MistralAI does it support timestamping? getting 404 on the docs

@cramforce when starting a session via slack it would give me the opencode web url, so to continue in the thread OR in opencode. the system still has lots of custom logic, probably Chat SDK can eliminate some of that

@cramforce A while ago, I built a custom slack agent based on a deployed OpenCode server. I built a mapping of one Slack thread <-> one OpenCode session <-> one sandbox env

Reasons why it makes sense to use chat-sdk.dev even if your agent only targets Slack: Lots of little details "just work": - Channel names are converted to clear text so your agent knows what people are talking about - Same with user names - Same in reverse: When the agent at-mentions somebody in clear text, that actually works - Link preview content, referenced posts, and images are automatically included in agent prompts, so that your agent has the full context - Agents love markdown, but Slack doesn't natively support it. Conversion from standard markdown to Slack-variant is automatic. - And that even works when using Slack's native append-only streaming API

@cramforce I can try to do that. I think the hard part the LLM (and I) might struggle with is to find a good abstraction for a "thread" object thats valid for a Slack (or Teams) thread as well as Claude/Codex/Opencode etc

@markwitt_me Honest question: Do you still need to do it now that you can just prompt it?

@cursor_ai I'd love a good prompt or subagent/skill that looks at active changes in a repo, then check recent conversation history and split into commits with smart messages