Terry Aney

@OrenMe @code Speaking of pricing. I have to make switch from Annual Pro to monthly. Is there a way to analyze my usage in terms of requests to see if I need to go to Pro+?

It’s getting clear as a bell 🛎️ @code page with clear token pricing across input/output/cache and vendors Full details in #pricing-tables" target="_blank" rel="nofollow noopener">docs.github.com/en/copilot/ref… It’s becoming apparent cost is now part of the equation and that might be a good thing against all odds(more on that later)

@bobstrogg @vkrajacic Airloom is equivalent to Copilot /remote?

@vkrajacic I’ve been building shitloads of stuff, but I’m too busy now with the number of concurrent projects 🫠 Here’s a couple though: sesh - Persistent remote shell sessions: github.com/BobStrogg/sesh Airloom - Remote shell from phone: github.com/BobStrogg/Airl…

X is flooded with people advocating for AI and vibe coding, but I see almost no cool demos or anything I can download to improve my daily work. I don't care how many tokens you burn, how many LOC you generate, which IDE you use. Can I use your application, please?

@burkeholland @_Evan_Boyle Tried my first cloud agent/issue. "You've reached your weekly rate limit. Please wait for limit to reset on April 27, 2026 at 12:00 AM or switch to auto model to continue." lol. Do I need to delete/redo PR to change to auto? I've tried everything.

@pierceboggan @digitarald Maybe @digitarald fixing with below, but FYI, I started a new convo like chat suggested and pasted the 'write plan', it said no memory, then I went back to the previous session (mode was still agent), and then I was able to do it again. So, switching in and out of chat did trick

@terryaney @digitarald could this be because when we swap modes the plan instructions are still around, confusing the model?

@pierceboggan On insider build Version: 1.118.0-insider 46e64ce993dba8618aaa24e7194639fc92147d7f. Started a new chat in plan mode. Then changed to agent and I've been 'arguing' for it to write a file. FYI

@burkeholland @coder Curious, haven't been following Max activity for a bit. Originally, you said "Max repo out there, but build your own"...is that still the modus operandi or are you now open to people just cloning and using what you've built? :) At least to play with it...

Added some big updates to Max agent over the weekend. * Container support (compose up) * Multi-model agents (@designer, @coder, @researcher, etc) * Memory rebuilt around Karpathy wiki gist burkeholland.github.io/max

@RickDrizin @SamsungUS Thanks. Will look, but don't know if I have a thermostat for the 'fridge' and/or 'freeze'. I have one for the 'soda drawer'. But going to look.

@terryaney @SamsungUS We just found out that by bumping freezer temperature from 0F to 1F fixed (unclogged) the ice maker. Hair drier alone wasn't working. So looks like a thermostat problem. Maybe it also works for you

Hey @SamsungUS I have your most expensive refrigerator, the ice maker broke under 3 years (it's a common problem, there's even a lawsuit for that issue) and your support is referring me to a service company that has only 1-star reviews. Should I throw it away and get a Whirlpool?

@davidfowl Curious, still grinding with Tally? Or did it serve its purpose and no longer needed? Had a todo to get fluent in how to control/use it and just wondering what your usage is now.

@pierceboggan @_Evan_Boyle @burkeholland Didn't mean feature parity (assumed that was true)... rather settings: paths: `%APPDATA%/Code/User/prompts/*.agent.md`(vscode) vs `~/.copilot/agents/*.agent.md` (cli) Agent model names: `GPT-5.4 (copilot)` (vscode) vs `gpt-5.4` (cli) Is that what you meant?

@terryaney @_Evan_Boyle @burkeholland Yes, of course! Our goal is we get to a consistent product experience for GitHub Copilot agent regardless of where you use it, including customizations.

@_Evan_Boyle @pierceboggan @burkeholland Curious. Been creating a skill for my team to auto install prompts, agents, skills, etc. that all use. Currently duplicating/massaging each for CLI vs Code. Is there plan to converge into one settings location/format in the future?

@eliostruyf @stanzillaz @JamesMontemagno Is there any reason to NOT run insiders? I've always only ran RTM...but I wanted to peek at this.

@stanzillaz @JamesMontemagno Correct

This new Visual Studio Code Agents app looks like my new #GitHub Copilot companion app. 😍 #VSCode

Maybe we are there since my last 'agent' addition...going to check, but before I continue to 'battle' with this skill, thought I'd ask.

@mariorod1 I feel like I enabled it (in cli via --experimental) but I have yet to see any rubber duck thoughts :(

Seeing more and more internal teams using rubber duck 🦆 and loving it. github.blog/ai-and-ml/gith…

@mkristensen Not as polished as yours, but had to try and vibe code a VS Code version of your Comment Studio ;) Works pretty good though. github.com/terryaney/Exte…

@burkeholland Doesn't anvil address this ... at least partially?

Exactly this. This is exactly the problem well stated. We gotta fix this at the root instead of expecting AI to magically work if we just hold it right. We don’t have the language or tooling for this yet. The stack is missing a layer.

Ah crap...to fast. There was really no visual notification. I restarted my terminal and still .17...then after tweet above and I went back to terminal, it finally was .18. Not sure if that is expected experience, but I'm updated :)

@_Evan_Boyle @burkeholland Where would I get logs/info to send along in issue (or tweet, lol) to help figure out why /update does nothing in my copilot cli?

@burkeholland @_Evan_Boyle Prompted with this, model Opus, agent anvil. And the response. So maybe expected, but I've been seeing stuff like this last few days.

@burkeholland @_Evan_Boyle Anything up with Copilot CLI in last day or two? I'm losing sessions. When I restart VS code and do /resume or /sessions the session I was JUST working on is no where to be found :(

@burkeholland anvil's about to be given a talking to...

We cannot continue to let AI write code the way we currently are. It’s way too good at cutting corners. I think hooks in the Copilot CLI might be the answer here. And I think we may only need two of them. preToolUse - the gate: Code gets linted or the model doesn't get to save the file. Model doesn't get to commit until all tests pass. No exceptions. The AI literally cannot move forward without passing basic muster. postToolUse - the feedback loop: Run your type checker after every edit (tsc --noEmit, whatevs). The output goes straight back into the model's context. It sees the errors. It fixes them. Loop restarts and continues until code is actually correct. No rationalizing about why it’s ok for some dumb error to persist. The model cannot write a file, pass a type check, or commit without actually doing what was asked. It’s time we held models to the same basic standards we hold for ourselves.