Michael Vessia

@ninan_phillip Both - but I wound up building github.com/MichaelVessia/… after posting this haha

@michaelvessia dmux might fit the bill. but a lot of coding agents are starting to support worktree's natively now. you a codex of claude guy? dmux.ai

Last night I tried like 5 agent/worktree management tools. Few are shaped as I'd like (terminal, vim bindings, using the existing agent CLIs so I can use my subscriptions). Sidecar checked these boxes but has way too many features imo. I get why everyone winds up building one

@pilcrowonpaper i think a lot more sites can afford to do this than they would like to admit and it would make some things quite simple that are otherwise very complicated ..

idk how to explain but I have a slight romanticization (?) of websites that can just periodically shutdown for a few hours for maintenance it's such a simple model

@aarondfrancis @0xblacklight My wife didn't realize she was signing up for a polycule with Claude and codex

@0xblacklight Future ex girlfriend

Sitting outside a coffee shop on Union street listening to some guy explain Claude code to his girlfriend and tell her about his setup and how it runs his emails

@alexhillman Please don't give them any ideas or I'll have to do a "someone who is good at the economy please help me budget this" post

through exactly every lens I can think of, anthropic could charge me 10X, and I’ve still spent more to get less.

@IanLandsman I think this is a smaller/faster model while opus fast mode is the same model with more compute

The amount of corporate espionage going on between OpenAI and Anthropic seems to be absolutely massive 😄 No other reasonable explanation for constantly shipping identical features within hours/days of each other.

@willmcgugan Very cool. I'm currently on a work subscription plan but I'll keep this on my radar if I wind up with an API key somehow. Nice work

@michaelvessia It is the Claude API under the hood. All above board.

What are you working on, Will? 🤷‍♂️

@craigjellick Using symlinks but it's an imperfect solution because the models respond differently to different prompting. We're heavily skewed towards anthropic models though so I haven't bothered solving that problem.

If a team wants to use both Claude and not-Claude, what’s the M.O. for CLAUDE .md vs AGENTS .md? Symlink? Maintain both?

@ethanniser Need this in my life

There are some people where after you talk to them everything seems to make sense and you wonder why you were ever so confused/scared/worried in the first place Love those people

@_brimtown All good! Gonna see if I can get my agents to send up a PR. Never wrote Go so we'll see how it comes out.. worst case it acts as a bug report for you

@michaelvessia sorry it sucked. we def need to sand the edges with like 50 agents in parallel trying some tasks + compiling a trip report

free alpha: early drop of a first party datadog cli github.com/DataDog/pup

@jessfraz I don't know how to talk to codex yet. Claude understands me

I have no idea why people would still be using Claude, Codex is so much better and it’s been like this for months since October

CPU temp at 88c, feelsgoodman

It's so cold here in NY i'm compiling rust programs to make my laptop act as a space heater

@astuyve @amorriscode The employers who advertised unlimited token budget as a perk are gonna start sweating 😅

@amorriscode Yeahhhh it’s good, glad I’m not the one paying for tokens right now though

literally can't live without fast mode

@dillon_mulroy @stolinski My clanker started hitting answer overflow today unprompted. Unfortunately it returned 429s but I still did the Leo pointing at the screen meme

when @stolinski randomly shows up in your clanker logs

@Austen It's so horrible at inferring my intent that I think I can only use it by handing it plans from another agent. Or I need to completely rewire my brain after months of strictly using Claude.

Slowly becoming Codex pilled

@dhasandev @_colemurray I haven't figured out how to get the slack integration from providers like anthropic to be able to use mcp / skills which would allow for access to useful systems

@_colemurray Now I’m curious why people wouldn’t just use droid/claude/codex or some provider from this. Do they not have that? Do your customers like the open source nature of it? I’m glad openinspect is moving fast

OpenInspect is moving fast! After a week of deployments with early customers, I've received a ton of feedback and iteration cycles. With that, new features have been added! Here's a recap:

@AmpCode @sqs @thorstenball Major respect for unshipping features. Never seen it done with such conviction.

Episode 10 of Raising An Agent with @sqs and @thorstenball is out! There's no better summary than this quote: "We will be killing our editor extension, the Amp VS Code extension. We're going to be killing it. And we're going to be killing it because we think it's no longer the future. We think the sidebar is dead. Let's walk through why." Topics in this episode: - The new deep mode in Amp - Balancing developer experience for humans & agents - Killing the VSCode extension & shift away from traditional editors - Pi & OpenClaw, two wonderful projects - Importance of reinventing yourself in AI Enjoy! And happy hacking! Timestamps: 01:00 Deep Mode 10:30 Optimizing the codebase for agents 15:00 Feature Preview: which Skills does your team use? 18:00 Balancing DX for humans & agents 21:35 Killing the Amp editor extension 28:00 The future of software and what it means 33:00 You need to stay agile 36:00 Pi & OpenClaw 39:00 Text editors holding companies back 44:00 Is manual context management coming to an end? 49:00 New concept for Threads 50:00 Amp, the business & the art installation