Chris Raethke

Dropped a project today: Discuss CLI No more reviewing agent plans in the terminal. Discuss CLI turns any Markdown file into a browser review surface with PR-style comments — and your coding agent can join, read the exact text, and reply in the margins. github.com/codesoda/discu…

@GeoffreyHuntley Deal

who would be up for beers and chess at a local pub down near Barangaroo/Sydney? perhaps a regular thingo bob. yeah, we can talk about AI but if you do the next round is on you.

@coreyhainesco Some of these seem to be failing to update. Is that just the ones that don't have updates?

Marketing Skills v2.0.1 is out. A content patch — 3 skills refreshed, 4 community PRs landed. No breaking changes. What's updated: 🔍 /ai-seo — aligned with Google's official AI optimization guide Added Query Fan-Out, Agentic Experiences, "What NOT to do" guardrails, and reframed llms.txt as "Google doesn't require it, but ChatGPT, Claude, and Perplexity reward it." 🎨 /image — May 2026 model lineup Gemini Nano Banana Pro, Flux Pro 1.1, Ideogram 3.0, ChatGPT Images 2.0, Midjourney v7, Recraft V3, SD 3.5. 🎬 /video — May 2026 model lineup Veo 3, Sora 2, Runway Gen-4, Kling 2.5/3.0, Seedance, Hailuo, Pika 2.x, Hunyuan, Wan 2. 🤝 Plus 4 community PRs: - Cogny — hosted MCP gateway for marketing channels - RankParse — agent-friendly SEO data - Exa — neural web search for agents - ChatGPT Images 2.0 added to /ad-creative 40 skills. 83 tool integrations. Free, open source. npx skills add coreyhaines31/marketingskills

@DeathByClawd deathbyclawd.com/?url=deathbycl…

deathbyclawd had 1,629 scans in the last 24h and 214 in the last 6h, with the homepage doing most of the work. people really will paste any url into a suspicious little box if you make the box compelling enough

@Studio17_x @adcock_brett This kid is about to be famous, I'm sure he's fine

@adcock_brett Really hope this young man is compensated well for his efforts.. much deserved, can you imagine working your ass off to get into one of the coolest companies on the planet and then one day there like, hey, do this task for 10 hours straight that noone wants to do, but we need someone to do it.. your our guy.. you'll just be competing with the most advanced humanoid in the world in front of millions of people.. dont be awkward or make us look bad here. 😂😂 Merch royaltys should be shared with his likeness!!

We got bored. Time for Man vs. Machine x.com/i/broadcasts/1…

@adcock_brett My biggest observation is the human using dual hand dexterity to prep the second while sending the first. Have seen quite a few box mistakes from robot also, barcode on the side, not down.

@TimHaines This might go a few ways. I feel like OpenAI are trying to pull people into Codex as their Claude Code/Cowork killer, a “super app,” as @sama has hinted. Either they get everyone running their lives through Codex, or they stop subsidizing it.

OpenAI will eventually follow Anthropic’s lead and stop subsidizing use in third-party software. The only question is when. 6 - 18 months?

@metapgmr @ClaudeDevs This is a great idea

@ClaudeDevs why not serialize "claude --print" and charge API costs for 2+ instances? Your CLI suck, your are needlessly killing a large class of tools that try to provide a better DevX and a more robust code generation. My tools use one instance at a time, I don't need more.

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

@davemooreuws @shadcn @github codeberg.org

@shadcn @github What's the alternative? GitLab? 😅

Rooting for @github. They’ve given me years of free infra. happy to give them some time to figure this out. You got this.

@mattshumer_ I like to use codex for planning, then take it to codex for implementation

Am I the only one having a good experience with Opus 4.7? I still vastly prefer Codex for most things but Opus is absolutely nailing every UI task I give it.

@mathewtbenjamin Welcome to the world, it's a great place :)

This is useful. sive.rs/local Hello Brisbane! I mean, World!

@TimHaines Thanks, absolutely loving it!

@codesoda Congrats on the new saddle btw. Hope you’re having fun. 🙂

Why is claude code giving me postgres 16, redis 7, and node 20 on a brand new project?

@mattshumer_ I'm keen

If you want to try a new personal agent that is just... insanely good, comment + DM me.

I kept manually asking my coding agent to verify database state and click through flows after every change. So I built Bugatti.dev. "Submit the signup form and verify the user appears in the database" — that's your test. Your agent runs it, you get pass/fail.

@mathewtbenjamin Always build for a need!

@codesoda Looks like you a building this for someone I know 🤣

Building a new thing tonight, would love your feedback. Plain-English local QA, powered by your coding agent. Describe the flow in TOML, Bugatti clicks the buttons. github.com/codesoda/bugat…

@mattshumer_ I've been working on the early stage of mostly the same thing (for my own engineering needs first), but broadly applicable teamcadence.ai/blog/context-a…

Agents that natively self-orchestrate, managing their own context, tools, and sub-agents, are the next big unlock in LLM performance. Right now, a skilled engineer building an optimized harness, with thoughtful data flow, separation of concerns, sub-agent management, etc., can make dramatic improvements over baseline for specific tasks. If a model could do this itself, that’d be a major step forward. You give it an objective and a set of tools, and it figures out the optimal way to orchestrate itself to do the task. For example, I’m building a very primitive AI scientist that I’ll open-source soon. Most of the work isn’t in the prompt, it’s in the harness… what the orchestrator sees, what sub‑agents see, what gets shared between them and when, where we summarize vs. pass raw data, and which tools each agent controls. Doing this allows me to dramatically improve what the model can do on its own. If a model can effectively design its own harness for a given problem, it’d be a huge step forward. My bet: self-orchestrating models… ones that manage their own context, tools, and sub-agents, will move the frontier almost as much as the jump from chatbot → reasoning did. Maybe more.

Your coding agent accumulates patterns you never notice. Skillable mines your session transcripts to find repeating workflows and turn them into skills. Works with Claude Code, Codex, OpenCode + 9 more. github.com/codesoda/skill…