I hated making this video...

This is one of those videos I really This is one of those videos I really don't want to make, but I've been having don't want to make, but I've been having don't want to make, but I've been having a lot of experiences that I wanted to a lot of experiences that I wanted to a lot of experiences that I wanted to share. As you've probably seen, I've share. As you've probably seen, I've share. As you've probably seen, I've been using Fable a lot. Sadly, Anthropic been using Fable a lot. Sadly, Anthropic been using Fable a lot. Sadly, Anthropic restricts the hell out of your restricts the hell out of your restricts the hell out of your subscriptions, so you kind of have to subscriptions, so you kind of have to subscriptions, so you kind of have to use it through their interfaces. In this use it through their interfaces. In this use it through their interfaces. In this case, Claude Code. I've been using case, Claude Code. I've been using case, Claude Code. I've been using Claude Code a lot the last couple days, Claude Code a lot the last couple days, Claude Code a lot the last couple days, probably as much as I used it back in probably as much as I used it back in probably as much as I used it back in December and January when I first got December and January when I first got December and January when I first got addicted to Opus 4.5. I've been through addicted to Opus 4.5. I've been through addicted to Opus 4.5. I've been through it with Claude Code, to put it lightly. it with Claude Code, to put it lightly. it with Claude Code, to put it lightly. And while there are lots of things I do And while there are lots of things I do And while there are lots of things I do not like about it, especially the not like about it, especially the not like about it, especially the desktop app, I have come around to a desktop app, I have come around to a desktop app, I have come around to a handful. And I've even found a bunch of handful. And I've even found a bunch of handful. And I've even found a bunch of small things that I think Claude Code small things that I think Claude Code small things that I think Claude Code does better than every other harness and does better than every other harness and does better than every other harness and CLI for doing agentic coding. I can't CLI for doing agentic coding. I can't CLI for doing agentic coding. I can't believe I'm doing this, but I think I believe I'm doing this, but I think I believe I'm doing this, but I think I have to. We're going to do a video about have to. We're going to do a video about have to. We're going to do a video about the good parts of Cloud Code. Think I the good parts of Cloud Code. Think I the good parts of Cloud Code. Think I need a hat for this one. Let's do this. And as a reminder, just Let's do this. And as a reminder, just because I have a hat and I have because I have a hat and I have because I have a hat and I have convinced a lot of people to spend a lot convinced a lot of people to spend a lot convinced a lot of people to spend a lot of money on Anthropic, they are not a of money on Anthropic, they are not a of money on Anthropic, they are not a sponsor. I don't think they ever would sponsor. I don't think they ever would sponsor. I don't think they ever would be considering some of the things I've be considering some of the things I've be considering some of the things I've said about this company in the past. But said about this company in the past. But said about this company in the past. But I do need to get paid, especially after I do need to get paid, especially after I do need to get paid, especially after the seven subscriptions that I've the seven subscriptions that I've the seven subscriptions that I've activated for myself and my team over activated for myself and my team over activated for myself and my team over the last few days. I'm $1,400 down, so the last few days. I'm $1,400 down, so the last few days. I'm $1,400 down, so please pardon me for getting a little please pardon me for getting a little please pardon me for getting a little bit of money from today's sponsor. Hey bit of money from today's sponsor. Hey bit of money from today's sponsor. Hey you, you're probably a real dev, right?

you, you're probably a real dev, right? you, you're probably a real dev, right? Like the type that uses Docker. Well, if Like the type that uses Docker. Well, if Like the type that uses Docker. Well, if you are, you're wasting a ton of time. you are, you're wasting a ton of time. you are, you're wasting a ton of time. Almost all Docker users are. At least Almost all Docker users are. At least Almost all Docker users are. At least all of the ones that aren't already all of the ones that aren't already all of the ones that aren't already using today's sponsor, Depot. These guys using today's sponsor, Depot. These guys using today's sponsor, Depot. These guys are the kings of Docker. They made it so are the kings of Docker. They made it so are the kings of Docker. They made it so much faster that it feels unbelievable. much faster that it feels unbelievable. much faster that it feels unbelievable. Up to 40 times faster for real world use Up to 40 times faster for real world use Up to 40 times faster for real world use cases. Their infrastructure builds cases. Their infrastructure builds cases. Their infrastructure builds containers remotely, works for Intel and containers remotely, works for Intel and containers remotely, works for Intel and ARM CPU, so you Mac people are going to ARM CPU, so you Mac people are going to ARM CPU, so you Mac people are going to be more than set. And they also built be more than set. And they also built be more than set. And they also built their own fine grained caching for the their own fine grained caching for the their own fine grained caching for the layers for your Docker images on their layers for your Docker images on their layers for your Docker images on their own CDN. What this means is that after own CDN. What this means is that after own CDN. What this means is that after one dev runs the build, everyone else one dev runs the build, everyone else one dev runs the build, everyone else just pulls the cache instead, making just pulls the cache instead, making just pulls the cache instead, making everything absurdly fast. And this isn't everything absurdly fast. And this isn't everything absurdly fast. And this isn't just good for running builds on your just good for running builds on your just good for running builds on your machine. This is incredible for agents machine. This is incredible for agents machine. This is incredible for agents that are doing work in the cloud, but that are doing work in the cloud, but that are doing work in the cloud, but more realistically, it's incredible for more realistically, it's incredible for more realistically, it's incredible for CI, which is why they provide GitHub CI CI, which is why they provide GitHub CI CI, which is why they provide GitHub CI that's up to 10 times faster than GitHub that's up to 10 times faster than GitHub that's up to 10 times faster than GitHub actions. If you're willing to move to actions. If you're willing to move to actions. If you're willing to move to something a little different from GitHub something a little different from GitHub something a little different from GitHub actions, their CI engine is so much actions, their CI engine is so much actions, their CI engine is so much better. Not only does it handle actual better. Not only does it handle actual better. Not only does it handle actual parallelism where it does one step like parallelism where it does one step like parallelism where it does one step like installs and then has three things installs and then has three things installs and then has three things running at the same time, which is running at the same time, which is running at the same time, which is unbelievably useful when you're doing unbelievably useful when you're doing unbelievably useful when you're doing real CI work, it also runs way faster on real CI work, it also runs way faster on real CI work, it also runs way faster on much better infrastructure. And most much better infrastructure. And most much better infrastructure. And most importantly, it's accessible via API and importantly, it's accessible via API and importantly, it's accessible via API and CLI. So your agents are able to run your CLI. So your agents are able to run your CLI. So your agents are able to run your CI in a loop without having to push up CI in a loop without having to push up CI in a loop without having to push up the changes. No more filing PRs just to the changes. No more filing PRs just to the changes. No more filing PRs just to trigger CI to see what works and then trigger CI to see what works and then trigger CI to see what works and then copy pasting the errors back to your copy pasting the errors back to your copy pasting the errors back to your agent. Just let the agent run. If this agent. Just let the agent run. If this agent. Just let the agent run. If this ad was longer than your Docker builds, ad was longer than your Docker builds, ad was longer than your Docker builds, get that fixed at soyv.link/devo.

get that fixed at soyv.link/devo. get that fixed at soyv.link/devo. So what the hell am I doing making a So what the hell am I doing making a So what the hell am I doing making a video about good parts of Cloud Code? video about good parts of Cloud Code? video about good parts of Cloud Code? Well, as I hinted at the start, I've Well, as I hinted at the start, I've Well, as I hinted at the start, I've been using it a ton, and I've noticed a been using it a ton, and I've noticed a been using it a ton, and I've noticed a handful of things I actually really handful of things I actually really handful of things I actually really like. from cool UI touches to nice like. from cool UI touches to nice like. from cool UI touches to nice features that only exist in Cloud Code features that only exist in Cloud Code features that only exist in Cloud Code to deep integrations with things like to deep integrations with things like to deep integrations with things like their skill management that are actually their skill management that are actually their skill management that are actually quite good. And I'm mostly making this quite good. And I'm mostly making this quite good. And I'm mostly making this video not to glaze Claude Code and video not to glaze Claude Code and video not to glaze Claude Code and convince everybody to go use it. I'll be convince everybody to go use it. I'll be convince everybody to go use it. I'll be transparent with my goal here. I want transparent with my goal here. I want transparent with my goal here. I want every other harness to steal all of every other harness to steal all of every other harness to steal all of these things. I am making this video not these things. I am making this video not these things. I am making this video not because Cloud Code is better and I want because Cloud Code is better and I want because Cloud Code is better and I want everyone to appreciate how great it is, everyone to appreciate how great it is, everyone to appreciate how great it is, but because I want everything else to but because I want everything else to but because I want everything else to take the good parts so that I don't miss take the good parts so that I don't miss take the good parts so that I don't miss Cloud Code when I use other things. So Cloud Code when I use other things. So Cloud Code when I use other things. So knowing that, what are those things? knowing that, what are those things? knowing that, what are those things? First and foremost, I want to talk about First and foremost, I want to talk about First and foremost, I want to talk about skills. You might be confused because skills. You might be confused because skills. You might be confused because skills are a feature in pretty much skills are a feature in pretty much skills are a feature in pretty much every one of these harnesses now. And every one of these harnesses now. And every one of these harnesses now. And they are. They're just markdown files, they are. They're just markdown files, they are. They're just markdown files, right? Mostly, yes. But this is where I right? Mostly, yes. But this is where I right? Mostly, yes. But this is where I start to really like what Claude Code start to really like what Claude Code start to really like what Claude Code does differently. One of the things does differently. One of the things does differently. One of the things Claude Code does really differently with Claude Code does really differently with Claude Code does really differently with their skills is the ability to execute their skills is the ability to execute their skills is the ability to execute scripts in them. This might sound scripts in them. This might sound scripts in them. This might sound insane, right? Like just tell the model insane, right? Like just tell the model insane, right? Like just tell the model to go execute the script. But hear me to go execute the script. But hear me to go execute the script. But hear me out. There's a lot of use cases where out. There's a lot of use cases where out. There's a lot of use cases where it's nice for the model to know things it's nice for the model to know things it's nice for the model to know things when it loads the skill instead of when it loads the skill instead of when it loads the skill instead of having to load the skill and then do having to load the skill and then do having to load the skill and then do things and then do the next step. Here's things and then do the next step. Here's things and then do the next step. Here's an example of a real skill I made that I an example of a real skill I made that I an example of a real skill I made that I use all the time. It's repo explorer.

use all the time. It's repo explorer. use all the time. It's repo explorer. This is my markdown file that replaced This is my markdown file that replaced This is my markdown file that replaced Ben's whole product that he was building Ben's whole product that he was building Ben's whole product that he was building that I was trying to convince him he that I was trying to convince him he that I was trying to convince him he doesn't need to build that product. doesn't need to build that product. doesn't need to build that product. Now he is convinced he understands, but Now he is convinced he understands, but Now he is convinced he understands, but at the time this was originally a demo at the time this was originally a demo at the time this was originally a demo and I've actually found myself using it and I've actually found myself using it and I've actually found myself using it all the time. The point of this skill is all the time. The point of this skill is all the time. The point of this skill is to allow agents to explore to allow agents to explore to allow agents to explore implementations of dependencies. Like if implementations of dependencies. Like if implementations of dependencies. Like if I'm using effect or react or some weird I'm using effect or react or some weird I'm using effect or react or some weird open- source project that the model open- source project that the model open- source project that the model doesn't know that much about, instead of doesn't know that much about, instead of doesn't know that much about, instead of just hunting through the docs and hoping just hunting through the docs and hoping just hunting through the docs and hoping it finds the right thing, this skill it finds the right thing, this skill it finds the right thing, this skill allows the model to clone the open allows the model to clone the open allows the model to clone the open source repo into a specific directory source repo into a specific directory source repo into a specific directory and then explore that specifically. It and then explore that specifically. It and then explore that specifically. It has a pretty simple description. Clone has a pretty simple description. Clone has a pretty simple description. Clone and inspect external repositories in a and inspect external repositories in a and inspect external repositories in a reusable local exploration cache. Use reusable local exploration cache. Use reusable local exploration cache. Use the skill when the user asks to explore, the skill when the user asks to explore, the skill when the user asks to explore, inspect, investigate, compare, or answer inspect, investigate, compare, or answer inspect, investigate, compare, or answer questions about a repo that may not questions about a repo that may not questions about a repo that may not already be in the current workspace. The already be in the current workspace. The already be in the current workspace. The actual skill itself is pretty simple. actual skill itself is pretty simple. actual skill itself is pretty simple. Use this skill to explore repos without Use this skill to explore repos without Use this skill to explore repos without cluttering the active workspace. I have cluttering the active workspace. I have cluttering the active workspace. I have the specific directory on my computer. the specific directory on my computer. the specific directory on my computer. In my home path, I haveexplore/repos, In my home path, I haveexplore/repos, In my home path, I haveexplore/repos, which is the local cache for repos that which is the local cache for repos that which is the local cache for repos that I can explore. You can see the cache I can explore. You can see the cache I can explore. You can see the cache contents by running this command. It contents by running this command. It contents by running this command. It tells you to list the current tells you to list the current tells you to list the current repositories before deciding which to repositories before deciding which to repositories before deciding which to use. I even call out that certain host use. I even call out that certain host use. I even call out that certain host support skill injection here. Otherwise, support skill injection here. Otherwise, support skill injection here. Otherwise, it should run this script to find what's it should run this script to find what's it should run this script to find what's there. Then it says check whether the there. Then it says check whether the there. Then it says check whether the repo is already there. If it's not repo is already there. If it's not repo is already there. If it's not present, clone it, then explore it.

present, clone it, then explore it. present, clone it, then explore it. Relatively simple. You might be seeing Relatively simple. You might be seeing Relatively simple. You might be seeing the first thing that cloud code does the first thing that cloud code does the first thing that cloud code does that every other harness doesn't. That's that every other harness doesn't. That's that every other harness doesn't. That's really nice. The current cache contents really nice. The current cache contents really nice. The current cache contents section. If the harness is capable, it section. If the harness is capable, it section. If the harness is capable, it will immediately make this directory if will immediately make this directory if will immediately make this directory if it doesn't already exist and then list it doesn't already exist and then list it doesn't already exist and then list its contents. This skips a whole step its contents. This skips a whole step its contents. This skips a whole step that the model has to do. And if every that the model has to do. And if every that the model has to do. And if every other harness supported this, I would be other harness supported this, I would be other harness supported this, I would be able to write the skill entirely able to write the skill entirely able to write the skill entirely differently. I would just open with the differently. I would just open with the differently. I would just open with the listing of the content. And then I would listing of the content. And then I would listing of the content. And then I would say right below, if the repo you want is say right below, if the repo you want is say right below, if the repo you want is in this directory, just go there. If in this directory, just go there. If in this directory, just go there. If not, clone it there. It would make this not, clone it there. It would make this not, clone it there. It would make this skill half the length and it would be skill half the length and it would be skill half the length and it would be meaningfully more reliable and meaningfully more reliable and meaningfully more reliable and resilient. But instead, I have to write resilient. But instead, I have to write resilient. But instead, I have to write it in two different ways. one where it in two different ways. one where it in two different ways. one where claude code gets these cool features and claude code gets these cool features and claude code gets these cool features and one where every other harness including one where every other harness including one where every other harness including codecs, pi and more don't. Thankfully, codecs, pi and more don't. Thankfully, codecs, pi and more don't. Thankfully, there's a pi extension that does this there's a pi extension that does this there's a pi extension that does this and I think open code is starting to and I think open code is starting to and I think open code is starting to introduce this as well. But this is a introduce this as well. But this is a introduce this as well. But this is a feature that quad code does really well feature that quad code does really well feature that quad code does really well that I think people are too hesitant to that I think people are too hesitant to that I think people are too hesitant to introduce. There are real risks with how introduce. There are real risks with how introduce. There are real risks with how much people just arbitrarily install much people just arbitrarily install much people just arbitrarily install skills that the skill could just skills that the skill could just skills that the skill could just arbitrarily execute code. We already arbitrarily execute code. We already arbitrarily execute code. We already deal with that with npm guys. Get over deal with that with npm guys. Get over deal with that with npm guys. Get over your [ __ ] selves. This is totally your [ __ ] selves. This is totally your [ __ ] selves. This is totally fine. If anything, it's better cuz the fine. If anything, it's better cuz the fine. If anything, it's better cuz the code is in here to be audited by an LLM.

code is in here to be audited by an LLM. code is in here to be audited by an LLM. It doesn't actually [ __ ] auto run. It doesn't actually [ __ ] auto run. It doesn't actually [ __ ] auto run. It's okay. It does kind of auto run once It's okay. It does kind of auto run once It's okay. It does kind of auto run once it's installed, but like you get the it's installed, but like you get the it's installed, but like you get the idea. This is not as unsafe as half the idea. This is not as unsafe as half the idea. This is not as unsafe as half the [ __ ] we do with npm. It's totally fine [ __ ] we do with npm. It's totally fine [ __ ] we do with npm. It's totally fine and this is not a new thing for me to and this is not a new thing for me to and this is not a new thing for me to call out. Two months ago, I already did call out. Two months ago, I already did call out. Two months ago, I already did call this out and I said that I want call this out and I said that I want call this out and I said that I want other things to be supportive of it like other things to be supportive of it like other things to be supportive of it like codec cli, pi, cursor, etc. I like this codec cli, pi, cursor, etc. I like this codec cli, pi, cursor, etc. I like this pattern a lot. It is genuinely really pattern a lot. It is genuinely really pattern a lot. It is genuinely really good and it's one of those things that good and it's one of those things that good and it's one of those things that should be part of the skills standard should be part of the skills standard should be part of the skills standard but apparently just isn't. Some part of but apparently just isn't. Some part of but apparently just isn't. Some part of me that's frustrated about anthropic me that's frustrated about anthropic me that's frustrated about anthropic wants to complain that they are moving wants to complain that they are moving wants to complain that they are moving too far away from the standard but I'm too far away from the standard but I'm too far away from the standard but I'm not going to because the standard should not going to because the standard should not going to because the standard should move. The standard should improve. We move. The standard should improve. We move. The standard should improve. We should take opportunities to do things should take opportunities to do things should take opportunities to do things better. And this is a thing that cla better. And this is a thing that cla better. And this is a thing that cla does better than every other harness does better than every other harness does better than every other harness right now. So we should all go embrace right now. So we should all go embrace right now. So we should all go embrace that. If for those looking for the pi that. If for those looking for the pi that. If for those looking for the pi skill, Joel made it. Joel hook/ pi skill skill, Joel made it. Joel hook/ pi skill skill, Joel made it. Joel hook/ pi skill interpolation. Joel's a legend. interpolation. Joel's a legend. interpolation. Joel's a legend. Apparently, our friend Mario did not Apparently, our friend Mario did not Apparently, our friend Mario did not like it though because it was not like it though because it was not like it though because it was not secure. So, know that as well going in, secure. So, know that as well going in, secure. So, know that as well going in, but that is the nature of this. It's but that is the nature of this. It's but that is the nature of this. It's arbitrary execution, but like it's it's arbitrary execution, but like it's it's arbitrary execution, but like it's it's not that big a deal. I think people are not that big a deal. I think people are not that big a deal. I think people are overreacting. And to their credit, overreacting. And to their credit, overreacting. And to their credit, Anthropic did also invent skills. So, I Anthropic did also invent skills. So, I Anthropic did also invent skills. So, I think it's totally fair for them to do think it's totally fair for them to do think it's totally fair for them to do stuff like this. I wish other harnesses stuff like this. I wish other harnesses stuff like this. I wish other harnesses would embrace the flexibility that would embrace the flexibility that would embrace the flexibility that Claude code skills have. Next, I want to Claude code skills have. Next, I want to Claude code skills have. Next, I want to talk about the Claude MD for a bit. I talk about the Claude MD for a bit. I talk about the Claude MD for a bit. I have issues here in particular that the have issues here in particular that the have issues here in particular that the Claude MD and agent MD are competing Claude MD and agent MD are competing Claude MD and agent MD are competing standards where everything uses the standards where everything uses the standards where everything uses the agent MD except for Claude which insists agent MD except for Claude which insists agent MD except for Claude which insists that it has a special file. I've come that it has a special file. I've come that it has a special file. I've come around to that a bit because Claude around to that a bit because Claude around to that a bit because Claude needs different instructions than other needs different instructions than other needs different instructions than other things do. It's annoying when I'm using

things do. It's annoying when I'm using things do. It's annoying when I'm using Claude models in open code or pi but I Claude models in open code or pi but I Claude models in open code or pi but I have come around to some of the cool have come around to some of the cool have come around to some of the cool things you could do in the cloud MD. I things you could do in the cloud MD. I things you could do in the cloud MD. I will highly recommend that you don't run will highly recommend that you don't run will highly recommend that you don't run their/init command. It's pretty bad. their/init command. It's pretty bad. their/init command. It's pretty bad. They have a new interactive version, but They have a new interactive version, but They have a new interactive version, but your Claude MD should be handwritten. your Claude MD should be handwritten. your Claude MD should be handwritten. The thing I want to call out that I The thing I want to call out that I The thing I want to call out that I think is genuinely really cool is think is genuinely really cool is think is genuinely really cool is importing additional files. Markdown importing additional files. Markdown importing additional files. Markdown doesn't have an import standard cuz doesn't have an import standard cuz doesn't have an import standard cuz Markdown's not really a standard. It's a Markdown's not really a standard. It's a Markdown's not really a standard. It's a vague vibe at best. So, Claude has a vague vibe at best. So, Claude has a vague vibe at best. So, Claude has a solution. Claude MD files can import solution. Claude MD files can import solution. Claude MD files can import additional files using at path/impport additional files using at path/impport additional files using at path/impport syntax. Imported files are expanded and syntax. Imported files are expanded and syntax. Imported files are expanded and loaded into context at launch alongside loaded into context at launch alongside loaded into context at launch alongside the Claude MD that references them. Both the Claude MD that references them. Both the Claude MD that references them. Both relative and absolute paths are allowed. relative and absolute paths are allowed. relative and absolute paths are allowed. Relative paths resolve relative to the Relative paths resolve relative to the Relative paths resolve relative to the file containing the import, not the file containing the import, not the file containing the import, not the working directory. Imported files can working directory. Imported files can working directory. Imported files can recursively import other files with a recursively import other files with a recursively import other files with a maximum depth of four hops. So if you maximum depth of four hops. So if you maximum depth of four hops. So if you want to intentionally include things want to intentionally include things want to intentionally include things like the readme package, JSON, etc., you like the readme package, JSON, etc., you like the readme package, JSON, etc., you can just tag them in. This also means can just tag them in. This also means can just tag them in. This also means you can do a very convenient thing for you can do a very convenient thing for you can do a very convenient thing for those of us who are annoyed that the those of us who are annoyed that the those of us who are annoyed that the agent MD and cloudmd are different.

agent MD and cloudmd are different. agent MD and cloudmd are different. Watch this. At agents.mmd. Watch this. At agents.mmd. Watch this. At agents.mmd. Now, my Claude MD is my agents MD. Not Now, my Claude MD is my agents MD. Not Now, my Claude MD is my agents MD. Not with a sim link, not with some weird with a sim link, not with some weird with a sim link, not with some weird [ __ ] It just is. But more [ __ ] It just is. But more [ __ ] It just is. But more importantly, if I do notice Claude doing importantly, if I do notice Claude doing importantly, if I do notice Claude doing some [ __ ] that my other models some [ __ ] that my other models some [ __ ] that my other models aren't, I can put additional stuff here aren't, I can put additional stuff here aren't, I can put additional stuff here just for Claude. So now I have the full just for Claude. So now I have the full just for Claude. So now I have the full content of my agent MD as well as my content of my agent MD as well as my content of my agent MD as well as my special Claude stuff just in the Cloud special Claude stuff just in the Cloud special Claude stuff just in the Cloud file without having to do any weird sim file without having to do any weird sim file without having to do any weird sim linking or managing two different files. linking or managing two different files. linking or managing two different files. It just it works. It's good. And if you It just it works. It's good. And if you It just it works. It's good. And if you have other docs in your repo that are have other docs in your repo that are have other docs in your repo that are useful to the model, you can tag that in useful to the model, you can tag that in useful to the model, you can tag that in here as well. This is a really nice here as well. This is a really nice here as well. This is a really nice pattern, especially because it works pattern, especially because it works pattern, especially because it works with things other than markdown. And I with things other than markdown. And I with things other than markdown. And I wish more things would copy it. The wish more things would copy it. The wish more things would copy it. The problem is that agents MD has become the problem is that agents MD has become the problem is that agents MD has become the standard. It's the thing everything standard. It's the thing everything standard. It's the thing everything uses. So if I do this in agents and it's uses. So if I do this in agents and it's uses. So if I do this in agents and it's supported by codecs, it might not be supported by codecs, it might not be supported by codecs, it might not be supported by five other things I use. So supported by five other things I use. So supported by five other things I use. So by having the cloudmom dbanthropic by having the cloudmom dbanthropic by having the cloudmom dbanthropic special thing, they're able to do stuff special thing, they're able to do stuff special thing, they're able to do stuff like this. They also have a nice like this. They also have a nice like this. They also have a nice override pattern that I actually quite override pattern that I actually quite override pattern that I actually quite like with claude.local.md.

like with claude.local.md. like with claude.local.md. If you add this to your global git If you add this to your global git If you add this to your global git ignore, which I would recommend. I'm ignore, which I would recommend. I'm ignore, which I would recommend. I'm probably going to go do that myself. It probably going to go do that myself. It probably going to go do that myself. It allows you to have specific instructions allows you to have specific instructions allows you to have specific instructions for how you want to work that will not for how you want to work that will not for how you want to work that will not affect the rest of your team. I do love affect the rest of your team. I do love affect the rest of your team. I do love that they have this as a separate thing that they have this as a separate thing that they have this as a separate thing that allows me to add my preferences that allows me to add my preferences that allows me to add my preferences without affecting you and how you work. without affecting you and how you work. without affecting you and how you work. Gets kind of rough in this way where Gets kind of rough in this way where Gets kind of rough in this way where when one person wants one cloud and when one person wants one cloud and when one person wants one cloud and somebody else wants another, git doesn't somebody else wants another, git doesn't somebody else wants another, git doesn't resolve that well. This does. This is a resolve that well. This does. This is a resolve that well. This does. This is a good thing that Anthropic has that as good thing that Anthropic has that as good thing that Anthropic has that as far as I know, none of the other far as I know, none of the other far as I know, none of the other harnesses do. I like this. While I was harnesses do. I like this. While I was harnesses do. I like this. While I was reading these docs to get this reading these docs to get this reading these docs to get this information, I found another really cool information, I found another really cool information, I found another really cool feature I didn't know about, which is feature I didn't know about, which is feature I didn't know about, which is launching sessions from links. They launching sessions from links. They launching sessions from links. They actually have their own HTTP code here, actually have their own HTTP code here, actually have their own HTTP code here, cloud- CLI slash, that allows you to cloud- CLI slash, that allows you to cloud- CLI slash, that allows you to open things in cloud CLI. So, you can open things in cloud CLI. So, you can open things in cloud CLI. So, you can have a page that someone makes that has have a page that someone makes that has have a page that someone makes that has a bunch of links in it to trigger cloud a bunch of links in it to trigger cloud a bunch of links in it to trigger cloud CLI. Apparently, Codex also has deep CLI. Apparently, Codex also has deep CLI. Apparently, Codex also has deep links, which is very, very nice. Cool to links, which is very, very nice. Cool to links, which is very, very nice. Cool to see. Does this work with app? The codeex see. Does this work with app? The codeex see. Does this work with app? The codeex app registers the codeex deep link. Oh, app registers the codeex deep link. Oh, app registers the codeex deep link. Oh, cool. That's actually really nice. That cool. That's actually really nice. That cool. That's actually really nice. That opens an app. Good [ __ ] I like that. I opens an app. Good [ __ ] I like that. I opens an app. Good [ __ ] I like that. I might even add this to my skill for might even add this to my skill for might even add this to my skill for making HTML plans where it has making HTML plans where it has making HTML plans where it has suggestions on how to kick off the job suggestions on how to kick off the job suggestions on how to kick off the job and I can click the cloud CLI link in and I can click the cloud CLI link in and I can click the cloud CLI link in that HTML page and it will just open up that HTML page and it will just open up that HTML page and it will just open up the terminal automatically. That's the terminal automatically. That's the terminal automatically. That's actually really cool. It does lead into actually really cool. It does lead into actually really cool. It does lead into the idea that your computer that you do the idea that your computer that you do the idea that your computer that you do work on is the same computer that your work on is the same computer that your work on is the same computer that your quad code runs on, which honestly hasn't quad code runs on, which honestly hasn't quad code runs on, which honestly hasn't been the best experience for me, even been the best experience for me, even been the best experience for me, even just for performance reasons. I've just for performance reasons. I've just for performance reasons. I've really enjoyed running quad code on really enjoyed running quad code on really enjoyed running quad code on other computers on my network, which other computers on my network, which other computers on my network, which this will not work for. But if you are

this will not work for. But if you are this will not work for. But if you are just using one computer with all of your just using one computer with all of your just using one computer with all of your stuff, this is actually a really cool stuff, this is actually a really cool stuff, this is actually a really cool pattern. If you combine this with HTML pattern. If you combine this with HTML pattern. If you combine this with HTML planning, you can make some really cool planning, you can make some really cool planning, you can make some really cool UX with this. It even lets you pass the UX with this. It even lets you pass the UX with this. It even lets you pass the working directory and a GitHub repo working directory and a GitHub repo working directory and a GitHub repo which will be resolved automatically by which will be resolved automatically by which will be resolved automatically by Claude Code if it's seen it before. If Claude Code if it's seen it before. If Claude Code if it's seen it before. If it doesn't, then it will make the it doesn't, then it will make the it doesn't, then it will make the session in your home directory instead session in your home directory instead session in your home directory instead interesting. So far, everything we've interesting. So far, everything we've interesting. So far, everything we've talked about isn't actually inside of talked about isn't actually inside of talked about isn't actually inside of Claude Code. It's just features it Claude Code. It's just features it Claude Code. It's just features it supports. But I want to hop into Cloud supports. But I want to hop into Cloud supports. But I want to hop into Cloud Code itself and show off some things Code itself and show off some things Code itself and show off some things that I actually find quite nice. In that I actually find quite nice. In that I actually find quite nice. In order to show the first one, I need to order to show the first one, I need to order to show the first one, I need to kick off some work. So, I'm going to do kick off some work. So, I'm going to do kick off some work. So, I'm going to do that. Do a quick audit of this work to that. Do a quick audit of this work to that. Do a quick audit of this work to make sure it's in a good state. then make sure it's in a good state. then make sure it's in a good state. then commit, push, and make a PR. Cool. So, I commit, push, and make a PR. Cool. So, I commit, push, and make a PR. Cool. So, I now have Mythos taking work it just did now have Mythos taking work it just did now have Mythos taking work it just did to go make a PR for it. But one of the to go make a PR for it. But one of the to go make a PR for it. But one of the really nice things Cloud Code has that I really nice things Cloud Code has that I really nice things Cloud Code has that I haven't seen anyone else do is a special haven't seen anyone else do is a special haven't seen anyone else do is a special command slash by the way, which lets you command slash by the way, which lets you command slash by the way, which lets you ask a quick side question without ask a quick side question without ask a quick side question without interrupting the main conversation. Can interrupting the main conversation. Can interrupting the main conversation. Can you tell me a bit about the performance you tell me a bit about the performance you tell me a bit about the performance or backups? And now I have a separate or backups? And now I have a separate or backups? And now I have a separate chat inside of this chat going while chat inside of this chat going while chat inside of this chat going while it's still doing the work that I told it it's still doing the work that I told it it's still doing the work that I told it to do. Apparently Codex has slashside to do. Apparently Codex has slashside to do. Apparently Codex has slashside which is really cool as well. I'm pumped which is really cool as well. I'm pumped which is really cool as well. I'm pumped that other things are doing this now.

that other things are doing this now. that other things are doing this now. But when I found this feature, it was But when I found this feature, it was But when I found this feature, it was the only one that had it. Super nice to the only one that had it. Super nice to the only one that had it. Super nice to just run a thing like this separately. just run a thing like this separately. just run a thing like this separately. Cool. And here we have the info on the Cool. And here we have the info on the Cool. And here we have the info on the performance. And while that was running, performance. And while that was running, performance. And while that was running, we got a bunch of sub aents queued up to we got a bunch of sub aents queued up to we got a bunch of sub aents queued up to explore. More cool things that Anthropic explore. More cool things that Anthropic explore. More cool things that Anthropic was first to, but a lot of other stuff was first to, but a lot of other stuff was first to, but a lot of other stuff now has, is the ability to like navigate now has, is the ability to like navigate now has, is the ability to like navigate through different sub aents it spun up. through different sub aents it spun up. through different sub aents it spun up. This isn't the coolest implementation of This isn't the coolest implementation of This isn't the coolest implementation of it, though. The coolest implementation it, though. The coolest implementation it, though. The coolest implementation by far is workflows, and nothing else by far is workflows, and nothing else by far is workflows, and nothing else has gone quite as far as Anthropic has has gone quite as far as Anthropic has has gone quite as far as Anthropic has here. One other little thing, and this here. One other little thing, and this here. One other little thing, and this is just one I've been using a lot is just one I've been using a lot is just one I've been using a lot because I've been stuck in the CLI more. because I've been stuck in the CLI more. because I've been stuck in the CLI more. Work trees are not my favorite way to Work trees are not my favorite way to Work trees are not my favorite way to deal with stuff, but for the ephemeral deal with stuff, but for the ephemeral deal with stuff, but for the ephemeral work where I spin up, do it, file PR, work where I spin up, do it, file PR, work where I spin up, do it, file PR, and then drop it forever. Having a and then drop it forever. Having a and then drop it forever. Having a d-work tree that actually behaves d-work tree that actually behaves d-work tree that actually behaves properly is really nice. Now I have the properly is really nice. Now I have the properly is really nice. Now I have the repo I was in, but in a subwork tree in repo I was in, but in a subwork tree in repo I was in, but in a subwork tree in the cloud/work trees, which has since the cloud/work trees, which has since the cloud/work trees, which has since been get ignored. This pattern's been get ignored. This pattern's been get ignored. This pattern's actually been very nice for me, and I've actually been very nice for me, and I've actually been very nice for me, and I've been enjoying it to just quickly spin up been enjoying it to just quickly spin up been enjoying it to just quickly spin up a work tree on some random idea I have. a work tree on some random idea I have. a work tree on some random idea I have. But what I want to show off is But what I want to show off is But what I want to show off is workflows. There's a couple ways to workflows. There's a couple ways to workflows. There's a couple ways to trigger a workflow. So the easiest is to trigger a workflow. So the easiest is to trigger a workflow. So the easiest is to just tell the model to do it. The most just tell the model to do it. The most just tell the model to do it. The most expensive and most reliable option expensive and most reliable option expensive and most reliable option though is Ultra Code. As grotesque as though is Ultra Code. As grotesque as though is Ultra Code. As grotesque as that purple gradient it does is I do that purple gradient it does is I do that purple gradient it does is I do often use it to maximize my utilization often use it to maximize my utilization often use it to maximize my utilization of my sub. That said, I'm trying now to of my sub. That said, I'm trying now to of my sub. That said, I'm trying now to use high and XH high directly more and use high and XH high directly more and use high and XH high directly more and just tell it to do a workflow. I'm going just tell it to do a workflow. I'm going just tell it to do a workflow. I'm going to voice the text because I'm lazy and to voice the text because I'm lazy and to voice the text because I'm lazy and want to just explain what I'm doing here want to just explain what I'm doing here want to just explain what I'm doing here at the same time. So let's do that. I at the same time. So let's do that. I at the same time. So let's do that. I want to audit the open PRs on this

want to audit the open PRs on this want to audit the open PRs on this project and figure out what their status project and figure out what their status project and figure out what their status is. Which ones have been trumped? Which is. Which ones have been trumped? Which is. Which ones have been trumped? Which ones don't need to be left open anymore ones don't need to be left open anymore ones don't need to be left open anymore because similar work has been merged. because similar work has been merged. because similar work has been merged. Which ones are ready to go? And which Which ones are ready to go? And which Which ones are ready to go? And which ones need a little more before they're ones need a little more before they're ones need a little more before they're worth pushing over the finish line. I worth pushing over the finish line. I worth pushing over the finish line. I want to use a workflow to break up all want to use a workflow to break up all want to use a workflow to break up all of this work. Before you run the of this work. Before you run the of this work. Before you run the workflow, please output the code you're workflow, please output the code you're workflow, please output the code you're going to use to run it so that we can going to use to run it so that we can going to use to run it so that we can read through it together. The last line read through it together. The last line read through it together. The last line of there is probably my favorite part of there is probably my favorite part of there is probably my favorite part because workflows aren't just a bunch of because workflows aren't just a bunch of because workflows aren't just a bunch of tool calls the agent can do. They fully tool calls the agent can do. They fully tool calls the agent can do. They fully embraced code mode, which is when you embraced code mode, which is when you embraced code mode, which is when you don't just tell the LLM to spit out text don't just tell the LLM to spit out text don't just tell the LLM to spit out text to trigger specific stuff and keep all to trigger specific stuff and keep all to trigger specific stuff and keep all of this stuff in its own memory. You let of this stuff in its own memory. You let of this stuff in its own memory. You let the agent write code to do a thing. the agent write code to do a thing. the agent write code to do a thing. Instead, instead of having all your MCP Instead, instead of having all your MCP Instead, instead of having all your MCP servers that flood you with thousands of servers that flood you with thousands of servers that flood you with thousands of rows from your database, you can just rows from your database, you can just rows from your database, you can just have it write code to call the MCP have it write code to call the MCP have it write code to call the MCP server, get those thousand rows, and server, get those thousand rows, and server, get those thousand rows, and then filter it programmatically before then filter it programmatically before then filter it programmatically before it ever makes its way into the context it ever makes its way into the context it ever makes its way into the context window. While this is running, I want to window. While this is running, I want to window. While this is running, I want to point out how I'm running Claude Code. point out how I'm running Claude Code. point out how I'm running Claude Code. because I'm not using the standard UI. I because I'm not using the standard UI. I because I'm not using the standard UI. I was an early mover to their full screen was an early mover to their full screen was an early mover to their full screen mode, which is using the alt screen mode, which is using the alt screen mode, which is using the alt screen rendering technique in the terminal. So rendering technique in the terminal. So rendering technique in the terminal. So instead of doing weird reaxes to replace instead of doing weird reaxes to replace instead of doing weird reaxes to replace text in the scroll back history and the text in the scroll back history and the text in the scroll back history and the buffer, which is how a lot of these buffer, which is how a lot of these buffer, which is how a lot of these terminal apps work, it instead takes terminal apps work, it instead takes terminal apps work, it instead takes over the full terminal. So when I scroll over the full terminal. So when I scroll over the full terminal. So when I scroll up, I can't scroll to before I ran cloud up, I can't scroll to before I ran cloud up, I can't scroll to before I ran cloud code. And when I close it, it fully code. And when I close it, it fully code. And when I close it, it fully closes and restores my old terminal closes and restores my old terminal closes and restores my old terminal view. So if I like, I don't know, lsa view. So if I like, I don't know, lsa view. So if I like, I don't know, lsa here, I have all this stuff. It's my here, I have all this stuff. It's my here, I have all this stuff. It's my custom command. so I don't leak. I run custom command. so I don't leak. I run custom command. so I don't leak. I run Claude here and now it's full screen and Claude here and now it's full screen and Claude here and now it's full screen and when I close it, I go back to the

when I close it, I go back to the when I close it, I go back to the terminal. That's cuz I'm using the full terminal. That's cuz I'm using the full terminal. That's cuz I'm using the full screen mode or the flickerless renderer. screen mode or the flickerless renderer. screen mode or the flickerless renderer. I found it pretty good. I would I found it pretty good. I would I found it pretty good. I would recommend switching to it if you're okay recommend switching to it if you're okay recommend switching to it if you're okay with not having traditional scroll with not having traditional scroll with not having traditional scroll buffer. It does get a little annoying buffer. It does get a little annoying buffer. It does get a little annoying when you're sshing with T-Max, but I when you're sshing with T-Max, but I when you're sshing with T-Max, but I just told my agent to go figure that out just told my agent to go figure that out just told my agent to go figure that out and fix it. And now I have fully working and fix it. And now I have fully working and fix it. And now I have fully working scroll in T-m over SSH, which has been a scroll in T-m over SSH, which has been a scroll in T-m over SSH, which has been a lifesaver. Apparently, you can turn this lifesaver. Apparently, you can turn this lifesaver. Apparently, you can turn this on by running slashtuy full screen in on by running slashtuy full screen in on by running slashtuy full screen in cloud code, but you can also do cloud cloud code, but you can also do cloud cloud code, but you can also do cloud code no flicker equals one as an code no flicker equals one as an code no flicker equals one as an environment variable. That's how I have environment variable. That's how I have environment variable. That's how I have it set. We now have the workflow it set. We now have the workflow it set. We now have the workflow written. So, here is what a workflow written. So, here is what a workflow written. So, here is what a workflow actually is. If you're not familiar, the actually is. If you're not familiar, the actually is. If you're not familiar, the point of a workflow is to take a bunch point of a workflow is to take a bunch point of a workflow is to take a bunch of work and break it up into groups of of work and break it up into groups of of work and break it up into groups of sub agents that have different system sub agents that have different system sub agents that have different system prompts, different roles, different prompts, different roles, different prompts, different roles, different things they are working on. It's not things they are working on. It's not things they are working on. It's not just like, okay, let's have six things just like, okay, let's have six things just like, okay, let's have six things go off and look at this stuff. It's a go off and look at this stuff. It's a go off and look at this stuff. It's a staged workflow that has different steps staged workflow that has different steps staged workflow that has different steps along the way. So, I was gonna tell it along the way. So, I was gonna tell it along the way. So, I was gonna tell it this looks good. Run it. Be cautious this looks good. Run it. Be cautious this looks good. Run it. Be cautious when you tell it to run a workflow when you tell it to run a workflow when you tell it to run a workflow though because these burn tokens. By though because these burn tokens. By though because these burn tokens. By default, workflows use up to eight default, workflows use up to eight default, workflows use up to eight agents in parallel and it will agents in parallel and it will agents in parallel and it will constantly be spinning up new ones as constantly be spinning up new ones as constantly be spinning up new ones as previous work is finished. Going to wait previous work is finished. Going to wait previous work is finished. Going to wait actually starts running so I can open up actually starts running so I can open up actually starts running so I can open up the workflow and show you. Here we are.

the workflow and show you. Here we are. the workflow and show you. Here we are. Once a workflow is running, you can go Once a workflow is running, you can go Once a workflow is running, you can go look at it here and it currently has 15 look at it here and it currently has 15 look at it here and it currently has 15 agents. it cued one for every single one agents. it cued one for every single one agents. it cued one for every single one of the PRs that is currently open of the PRs that is currently open of the PRs that is currently open because that's what I asked it to do to because that's what I asked it to do to because that's what I asked it to do to audit the PRs. Each time it audits one audit the PRs. Each time it audits one audit the PRs. Each time it audits one of those PRs, it will decide whether or of those PRs, it will decide whether or of those PRs, it will decide whether or not it needs to throw that PR over to not it needs to throw that PR over to not it needs to throw that PR over to this new rule phase and then after this new rule phase and then after this new rule phase and then after there's a third verify phase. It doesn't there's a third verify phase. It doesn't there's a third verify phase. It doesn't even necessarily go through the phases even necessarily go through the phases even necessarily go through the phases in order. It just sets up these sub in order. It just sets up these sub in order. It just sets up these sub agents to find some information and then agents to find some information and then agents to find some information and then decide if it should go through decide if it should go through decide if it should go through additional phases or not. It's a fully additional phases or not. It's a fully additional phases or not. It's a fully dynamic step-by-step workflow based on dynamic step-by-step workflow based on dynamic step-by-step workflow based on what you're asking for. It does burn what you're asking for. It does burn what you're asking for. It does burn usage like [ __ ] mad, though, so know usage like [ __ ] mad, though, so know usage like [ __ ] mad, though, so know that going in. So, we're at 9% used that going in. So, we're at 9% used that going in. So, we're at 9% used right now, and in not much time, that right now, and in not much time, that right now, and in not much time, that will bump. Yep, it's already up to 10. will bump. Yep, it's already up to 10. will bump. Yep, it's already up to 10. And if I refresh again in a minute, And if I refresh again in a minute, And if I refresh again in a minute, it'll be at like 12 or 14. From my it'll be at like 12 or 14. From my it'll be at like 12 or 14. From my testing, when I had a workflow going, testing, when I had a workflow going, testing, when I had a workflow going, just one workflow with those eight like just one workflow with those eight like just one workflow with those eight like parallel threads, it was costing about parallel threads, it was costing about parallel threads, it was costing about $100 every 10 minutes. Yeah. No. going $100 every 10 minutes. Yeah. No. going $100 every 10 minutes. Yeah. No. going in workflows are expensive with Fable in workflows are expensive with Fable in workflows are expensive with Fable specifically, but if you use them with specifically, but if you use them with specifically, but if you use them with cheaper models or you even tell Fable to cheaper models or you even tell Fable to cheaper models or you even tell Fable to orchestrate using Sonnet and Opus orchestrate using Sonnet and Opus orchestrate using Sonnet and Opus instead, you can make the usage here way instead, you can make the usage here way instead, you can make the usage here way less aggressive. But just those two less aggressive. But just those two less aggressive. But just those two sentences I said there were enough for sentences I said there were enough for sentences I said there were enough for it to go up to 12%. So use with caution.

it to go up to 12%. So use with caution. it to go up to 12%. So use with caution. But I want to read through the code it But I want to read through the code it But I want to read through the code it wrote for the workflow because one of wrote for the workflow because one of wrote for the workflow because one of the things I think is coolest about it. the things I think is coolest about it. the things I think is coolest about it. This is going to sound contradictory, so This is going to sound contradictory, so This is going to sound contradictory, so hear me out. You can never be more hear me out. You can never be more hear me out. You can never be more dynamic than code. I know that sounds dynamic than code. I know that sounds dynamic than code. I know that sounds crazy cuz code is code. It does what it crazy cuz code is code. It does what it crazy cuz code is code. It does what it says. But when the agent can write code, says. But when the agent can write code, says. But when the agent can write code, it is effectively building its own it is effectively building its own it is effectively building its own custom feature every time it does a custom feature every time it does a custom feature every time it does a workflow. If we were to build workflows workflow. If we were to build workflows workflow. If we were to build workflows as like a core feature that the agent as like a core feature that the agent as like a core feature that the agent just calls, it would have to be just calls, it would have to be just calls, it would have to be structured such that you tell the model, structured such that you tell the model, structured such that you tell the model, okay, here are the different phases you okay, here are the different phases you okay, here are the different phases you can do. Here's how you structure it. can do. Here's how you structure it. can do. Here's how you structure it. Here's where you put prompts and we'll Here's where you put prompts and we'll Here's where you put prompts and we'll go spin up the rest. But when you tell go spin up the rest. But when you tell go spin up the rest. But when you tell it to write code, it's now able to it to write code, it's now able to it to write code, it's now able to structure it however the [ __ ] it wants. structure it however the [ __ ] it wants. structure it however the [ __ ] it wants. And it takes advantage of that. I've And it takes advantage of that. I've And it takes advantage of that. I've seen Fable write workflows that just seen Fable write workflows that just seen Fable write workflows that just weren't compatible. Some of them weren't weren't compatible. Some of them weren't weren't compatible. Some of them weren't even valid JS because it's trying so even valid JS because it's trying so even valid JS because it's trying so hard to push the limits of what this hard to push the limits of what this hard to push the limits of what this feature can do which I do actually think feature can do which I do actually think feature can do which I do actually think is kind of cool. So it starts with this is kind of cool. So it starts with this is kind of cool. So it starts with this meta export which is the name of the meta export which is the name of the meta export which is the name of the workflow open PR audit says what it's workflow open PR audit says what it's workflow open PR audit says what it's going to do. Audit all 15 open PRs and going to do. Audit all 15 open PRs and going to do. Audit all 15 open PRs and it has three phases it defines. One is it has three phases it defines. One is it has three phases it defines. One is audit which is detailed as one readonly audit which is detailed as one readonly audit which is detailed as one readonly agent per PR. Then the rule is to pick a agent per PR. Then the rule is to pick a agent per PR. Then the rule is to pick a winner inside of each overlapping winner inside of each overlapping winner inside of each overlapping cluster and then verify, which is one cluster and then verify, which is one cluster and then verify, which is one adversarial checker per disposition. So adversarial checker per disposition. So adversarial checker per disposition. So every time it finds a thing it's not every time it finds a thing it's not every time it finds a thing it's not sure about, it will verify that here.

sure about, it will verify that here. sure about, it will verify that here. And here are all of the PRs. This is And here are all of the PRs. This is And here are all of the PRs. This is code that isn't actually being used code that isn't actually being used code that isn't actually being used directly for the workflow. It's not directly for the workflow. It's not directly for the workflow. It's not exported. It just has this here for its exported. It just has this here for its exported. It just has this here for its own usage. This is all info it got from own usage. This is all info it got from own usage. This is all info it got from probably running the GitHub CLI, getting probably running the GitHub CLI, getting probably running the GitHub CLI, getting this data, and then putting it into this this data, and then putting it into this this data, and then putting it into this JS file. Has each PR here. Here it says JS file. Has each PR here. Here it says JS file. Has each PR here. Here it says what the title is, whether or not it's a what the title is, whether or not it's a what the title is, whether or not it's a draft, whether or not it's mergeable, draft, whether or not it's mergeable, draft, whether or not it's mergeable, and when it was updated. Whether or not and when it was updated. Whether or not and when it was updated. Whether or not it's mergeable is like if it's it's mergeable is like if it's it's mergeable is like if it's conflicting or not, which is cool that conflicting or not, which is cool that conflicting or not, which is cool that it got that data ahead of time, so the it got that data ahead of time, so the it got that data ahead of time, so the agents aren't stuck figuring that out agents aren't stuck figuring that out agents aren't stuck figuring that out for each of themselves. Then we have by for each of themselves. Then we have by for each of themselves. Then we have by number, which is a mapping of all of the number, which is a mapping of all of the number, which is a mapping of all of the data here so that it's an object that is data here so that it's an object that is data here so that it's an object that is number, key, data, value, just to make number, key, data, value, just to make number, key, data, value, just to make it easier to go through. Then it it easier to go through. Then it it easier to go through. Then it actually pulled some stuff from memory actually pulled some stuff from memory actually pulled some stuff from memory because I've done this specific audit because I've done this specific audit because I've done this specific audit before. So here it had prior where PR35 before. So here it had prior where PR35 before. So here it had prior where PR35 won in a tournament against 37 and 39. won in a tournament against 37 and 39. won in a tournament against 37 and 39. PR37 lost that same tournament. PR39 PR37 lost that same tournament. PR39 PR37 lost that same tournament. PR39 superseded 35 and 37, but it was superseded 35 and 37, but it was superseded 35 and 37, but it was empirically refuted. All caps refuted. empirically refuted. All caps refuted. empirically refuted. All caps refuted. Chamra reference MD documents a return Chamra reference MD documents a return Chamra reference MD documents a return type that code doesn't produce. Two type that code doesn't produce. Two type that code doesn't produce. Two disjointing storage runtimes. Four or disjointing storage runtimes. Four or disjointing storage runtimes. Four or five smoke functions dead. Hosted. Yeah, five smoke functions dead. Hosted. Yeah, five smoke functions dead. Hosted. Yeah, it really did not like that PR. That PR it really did not like that PR. That PR it really did not like that PR. That PR was written by 55. So I fable. It's was written by 55. So I fable. It's was written by 55. So I fable. It's funny. Opus48 glazed the [ __ ] out of GPT funny. Opus48 glazed the [ __ ] out of GPT funny. Opus48 glazed the [ __ ] out of GPT code. Fable hates GPT code. It's code. Fable hates GPT code. It's code. Fable hates GPT code. It's actually really interesting how actually really interesting how actually really interesting how different they are in this way. I would different they are in this way. I would different they are in this way. I would say that this is probably the single say that this is probably the single say that this is probably the single area I've seen Fable differ the most area I've seen Fable differ the most area I've seen Fable differ the most from Opus. Opus loved OpenAI code. Fable from Opus. Opus loved OpenAI code. Fable from Opus. Opus loved OpenAI code. Fable hates it. And then it has the main logs.

hates it. And then it has the main logs. hates it. And then it has the main logs. This is things that have recently This is things that have recently This is things that have recently merged. So it has as a reference. Then merged. So it has as a reference. Then merged. So it has as a reference. Then it is best to group all the different it is best to group all the different it is best to group all the different clusters of PRs that are related. So all clusters of PRs that are related. So all clusters of PRs that are related. So all of these have to do with files. All of of these have to do with files. All of of these have to do with files. All of these have to do with whitelisting. All these have to do with whitelisting. All these have to do with whitelisting. All of these are off brokers. The rest are of these are off brokers. The rest are of these are off brokers. The rest are all unique. Cool that they did that for all unique. Cool that they did that for all unique. Cool that they did that for me. Dispositions. These are all the me. Dispositions. These are all the me. Dispositions. These are all the different states it can be in. The audit different states it can be in. The audit different states it can be in. The audit schema. This is how it wants the audit schema. This is how it wants the audit schema. This is how it wants the audit to return when the model is done. The to return when the model is done. The to return when the model is done. The ruling schema. This is for the ruler ruling schema. This is for the ruler ruling schema. This is for the ruler verifier layer. And then we have the verifier layer. And then we have the verifier layer. And then we have the verdict schema. The audit prompt which verdict schema. The audit prompt which verdict schema. The audit prompt which is the prompt that is dynamically is the prompt that is dynamically is the prompt that is dynamically generated as a function that takes in a generated as a function that takes in a generated as a function that takes in a PR. It also uses the today definition so PR. It also uses the today definition so PR. It also uses the today definition so that the model knows what today is that the model knows what today is that the model knows what today is without having to potentially destroy without having to potentially destroy without having to potentially destroy its context by putting that in the its context by putting that in the its context by putting that in the system prompt. It's actually I haven't system prompt. It's actually I haven't system prompt. It's actually I haven't read one of these in full before. This read one of these in full before. This read one of these in full before. This is super cool. This whole thing here is super cool. This whole thing here is super cool. This whole thing here actually is just a function that returns actually is just a function that returns actually is just a function that returns a string. It's a template string a string. It's a template string a string. It's a template string generator that you pass a PR to that has generator that you pass a PR to that has generator that you pass a PR to that has number, title, branch, all of this stuff number, title, branch, all of this stuff number, title, branch, all of this stuff to give that to the model as a prompt to give that to the model as a prompt to give that to the model as a prompt because again it's doing all of this because again it's doing all of this because again it's doing all of this dynamically on the fly which is super dynamically on the fly which is super dynamically on the fly which is super cool. Here it even has a dynamic turnary cool. Here it even has a dynamic turnary cool. Here it even has a dynamic turnary where it puts different context based on where it puts different context based on where it puts different context based on whether or not there are priors for this whether or not there are priors for this whether or not there are priors for this PR. Has a different prompt for ruling PR. Has a different prompt for ruling PR. Has a different prompt for ruling which is deciding whether or not this which is deciding whether or not this which is deciding whether or not this thing should be used or not. And then a thing should be used or not. And then a thing should be used or not. And then a verify prompt. You're an adversarial verify prompt. You're an adversarial verify prompt. You're an adversarial verifier in the lake bed repo. Read verifier in the lake bed repo. Read verifier in the lake bed repo. Read only. Origin already fetched. Do not only. Origin already fetched. Do not only. Origin already fetched. Do not fetch, check out, or modify. It knows fetch, check out, or modify. It knows fetch, check out, or modify. It knows how to prompt Claude. Historically, I how to prompt Claude. Historically, I how to prompt Claude. Historically, I didn't feel like models knew how to didn't feel like models knew how to didn't feel like models knew how to prompt themselves. Fable has pushed me prompt themselves. Fable has pushed me prompt themselves. Fable has pushed me way over the line here. Fable is great way over the line here. Fable is great way over the line here. Fable is great at prompting anthropic models. It knows at prompting anthropic models. It knows at prompting anthropic models. It knows all the ways they screw up and tells all the ways they screw up and tells all the ways they screw up and tells them not to ahead of time. The proposed them not to ahead of time. The proposed them not to ahead of time. The proposed disposition for PR is the data that it

disposition for PR is the data that it disposition for PR is the data that it fetched from the other agents. Try to fetched from the other agents. Try to fetched from the other agents. Try to refute this using actual repo state. the refute this using actual repo state. the refute this using actual repo state. the GitHub PR view diff checks git logs GitHub PR view diff checks git logs GitHub PR view diff checks git logs close superseded close stale yada yada close superseded close stale yada yada close superseded close stale yada yada it's a decent prompt and then we have it's a decent prompt and then we have it's a decent prompt and then we have the results where they call pipeline the results where they call pipeline the results where they call pipeline which is their magic global helper as which is their magic global helper as which is their magic global helper as clusters which are the groups that we clusters which are the groups that we clusters which are the groups that we talked about before an async function talked about before an async function talked about before an async function for uh auditing every PR member that's for uh auditing every PR member that's for uh auditing every PR member that's the first stage then stage two is the the first stage then stage two is the the first stage then stage two is the next section stage three is the end part next section stage three is the end part next section stage three is the end part but you define this all in a pipeline but you define this all in a pipeline but you define this all in a pipeline that takes the clusters which is the that takes the clusters which is the that takes the clusters which is the data as the first argument and then the data as the first argument and then the data as the first argument and then the functions to define the sub jobs and the functions to define the sub jobs and the functions to define the sub jobs and the subruns for all of these other sub subruns for all of these other sub subruns for all of these other sub aents. It's [ __ ] cool. It wrote 240 aents. It's [ __ ] cool. It wrote 240 aents. It's [ __ ] cool. It wrote 240 lines of code that is entirely throwaway lines of code that is entirely throwaway lines of code that is entirely throwaway that is only going to be executed ever that is only going to be executed ever that is only going to be executed ever once to trigger this workflow. That's once to trigger this workflow. That's once to trigger this workflow. That's [ __ ] awesome. Although I will admit [ __ ] awesome. Although I will admit [ __ ] awesome. Although I will admit with Fable that is also expensive. We with Fable that is also expensive. We with Fable that is also expensive. We are now up to 24% used in the time that are now up to 24% used in the time that are now up to 24% used in the time that I read that. Another 12% of my $200 a I read that. Another 12% of my $200 a I read that. Another 12% of my $200 a month 5 hour window. When you see month 5 hour window. When you see month 5 hour window. When you see everybody talking about agent loops and everybody talking about agent loops and everybody talking about agent loops and letting the agent prompt itself, this is letting the agent prompt itself, this is letting the agent prompt itself, this is a phenomenal example of it. Should you a phenomenal example of it. Should you a phenomenal example of it. Should you go use this and burn all of your usage go use this and burn all of your usage go use this and burn all of your usage yourself? Maybe. But at the very least, yourself? Maybe. But at the very least, yourself? Maybe. But at the very least, you should look at this and learn from you should look at this and learn from you should look at this and learn from it, realizing that this is where you get it, realizing that this is where you get it, realizing that this is where you get really crazy powers with models when you really crazy powers with models when you really crazy powers with models when you let them write their own code to do let them write their own code to do let them write their own code to do additional subprompting where code isn't additional subprompting where code isn't additional subprompting where code isn't just the output of a model. Code is a just the output of a model. Code is a just the output of a model. Code is a step between model runs. Genuinely step between model runs. Genuinely step between model runs. Genuinely really cool. and has got me thinking a really cool. and has got me thinking a really cool. and has got me thinking a lot more deeply about how to use these

lot more deeply about how to use these lot more deeply about how to use these things as well if I'm being honest with things as well if I'm being honest with things as well if I'm being honest with y'all. On the topic of usage limits y'all. On the topic of usage limits y'all. On the topic of usage limits though, I'm burning through them pretty though, I'm burning through them pretty though, I'm burning through them pretty fast. That's why I now have to use fast. That's why I now have to use fast. That's why I now have to use multiple accounts. Let's say I want to multiple accounts. Let's say I want to multiple accounts. Let's say I want to reduce my usage on this account right reduce my usage on this account right reduce my usage on this account right now. If I was near the end and I was now. If I was near the end and I was now. If I was near the end and I was scared, I'd have a lot of reasons I'd scared, I'd have a lot of reasons I'd scared, I'd have a lot of reasons I'd want to do that. We have this workflow want to do that. We have this workflow want to do that. We have this workflow already running. So, it's going to keep already running. So, it's going to keep already running. So, it's going to keep going. But a problem I've had is that going. But a problem I've had is that going. But a problem I've had is that when your usage limit gets hit, this run when your usage limit gets hit, this run when your usage limit gets hit, this run can stop and it's really hard to recover can stop and it's really hard to recover can stop and it's really hard to recover it, if possible at all. But what you can it, if possible at all. But what you can it, if possible at all. But what you can do to handle this is make sure you do to handle this is make sure you do to handle this is make sure you switch accounts before you get there. switch accounts before you get there. switch accounts before you get there. But Theo, isn't that going to screw up But Theo, isn't that going to screw up But Theo, isn't that going to screw up the run? Isn't that going to like break the run? Isn't that going to like break the run? Isn't that going to like break your current stuff? Not for my your current stuff? Not for my your current stuff? Not for my experience. As I mentioned, I have that experience. As I mentioned, I have that experience. As I mentioned, I have that workflow running right now. And I'm workflow running right now. And I'm workflow running right now. And I'm going to switch accounts. I have to hide going to switch accounts. I have to hide going to switch accounts. I have to hide myself doing this, but I'll show you. myself doing this, but I'll show you. myself doing this, but I'll show you. Going to click authorize. Login Going to click authorize. Login Going to click authorize. Login successful. And again, this agent was successful. And again, this agent was successful. And again, this agent was already running, but now all of its already running, but now all of its already running, but now all of its usage is going to be routed to this usage is going to be routed to this usage is going to be routed to this other account. So even though I already other account. So even though I already other account. So even though I already had a run going, it will start future had a run going, it will start future had a run going, it will start future turns on the other accounts off because turns on the other accounts off because turns on the other accounts off because every time it does a tool call, a new every time it does a tool call, a new every time it does a tool call, a new API request is made and once that API request is made and once that API request is made and once that happens, the new O is used. So it will happens, the new O is used. So it will happens, the new O is used. So it will keep using this in a given generation, keep using this in a given generation, keep using this in a given generation, but when it runs another command, it but when it runs another command, it but when it runs another command, it reads additional data, it does anything reads additional data, it does anything reads additional data, it does anything that requires execution of code, a new that requires execution of code, a new that requires execution of code, a new API request will be made and that one API request will be made and that one API request will be made and that one will go through your other off. This is will go through your other off. This is will go through your other off. This is awesome and really abusable. While I was awesome and really abusable. While I was awesome and really abusable. While I was recording this, a fun announcement went recording this, a fun announcement went recording this, a fun announcement went live that apparently uh Anthropic is no live that apparently uh Anthropic is no live that apparently uh Anthropic is no longer letting anyone use Fable because

longer letting anyone use Fable because longer letting anyone use Fable because the US government told them to suspend the US government told them to suspend the US government told them to suspend Fable and Mythos access for foreign Fable and Mythos access for foreign Fable and Mythos access for foreign actors. That is unprecedented. actors. That is unprecedented. actors. That is unprecedented. I have a couple more features I want to I have a couple more features I want to I have a couple more features I want to show off really quick before I forget. show off really quick before I forget. show off really quick before I forget. One that a lot of other things have, but One that a lot of other things have, but One that a lot of other things have, but I've actually found myself liking the I've actually found myself liking the I've actually found myself liking the implementation in Cloud Code quite a implementation in Cloud Code quite a implementation in Cloud Code quite a bit, is branch as well as rewind. Rewind bit, is branch as well as rewind. Rewind bit, is branch as well as rewind. Rewind I actually use a ton because if I I actually use a ton because if I I actually use a ton because if I accidentally send a prompt or get accidentally send a prompt or get accidentally send a prompt or get something wrong, I will use it to like something wrong, I will use it to like something wrong, I will use it to like go back one or two steps. I do have an go back one or two steps. I do have an go back one or two steps. I do have an issue with it, which is that if you do issue with it, which is that if you do issue with it, which is that if you do this erroneously, fastforwarding is this erroneously, fastforwarding is this erroneously, fastforwarding is nearly impossible. nearly impossible. nearly impossible. So be careful with rewind, but when you So be careful with rewind, but when you So be careful with rewind, but when you accidentally send something and want to accidentally send something and want to accidentally send something and want to go back, it is very nice. Slanch is the go back, it is very nice. Slanch is the go back, it is very nice. Slanch is the better option for when you want to go do better option for when you want to go do better option for when you want to go do something else or take the history you something else or take the history you something else or take the history you have now and go somewhere else with it. have now and go somewhere else with it. have now and go somewhere else with it. It's very, very nice. Thankful more It's very, very nice. Thankful more It's very, very nice. Thankful more things are starting to add that. I things are starting to add that. I things are starting to add that. I definitely recommend it. One more that a definitely recommend it. One more that a definitely recommend it. One more that a lot of people told me to talk about lot of people told me to talk about lot of people told me to talk about which I haven't used yet but I've heard which I haven't used yet but I've heard which I haven't used yet but I've heard really good things so I'm going to try really good things so I'm going to try really good things so I'm going to try it here is remote control. It's a it here is remote control. It's a it here is remote control. It's a builtin feature where I can type / builtin feature where I can type / builtin feature where I can type / remote control. So now I've given the remote control. So now I've given the remote control. So now I've given the ability for this cloud code instance on ability for this cloud code instance on ability for this cloud code instance on my real computer to be controlled my real computer to be controlled my real computer to be controlled through cloud.ai the website or from my through cloud.ai the website or from my through cloud.ai the website or from my phone using the claude app. Yeah, I just phone using the claude app. Yeah, I just phone using the claude app. Yeah, I just went to claude on my phone. I had to go went to claude on my phone. I had to go went to claude on my phone. I had to go to claude code of course but here you to claude code of course but here you to claude code of course but here you probably can't see great and I'm not probably can't see great and I'm not probably can't see great and I'm not going to do anything to fix that. I can going to do anything to fix that. I can going to do anything to fix that. I can click audit open PRs and workflow click audit open PRs and workflow click audit open PRs and workflow planning. And this is the exact run that planning. And this is the exact run that planning. And this is the exact run that is going on on my computer right now.

is going on on my computer right now. is going on on my computer right now. And it just hit a Python decoding error And it just hit a Python decoding error And it just hit a Python decoding error with some of the JSON that it got back with some of the JSON that it got back with some of the JSON that it got back as a response. Fascinating. Cool though. as a response. Fascinating. Cool though. as a response. Fascinating. Cool though. I have that on my phone now. I actually I have that on my phone now. I actually I have that on my phone now. I actually do really like that. I will probably use do really like that. I will probably use do really like that. I will probably use that more when I'm like running a bunch that more when I'm like running a bunch that more when I'm like running a bunch of [ __ ] and then out on the go. I wish of [ __ ] and then out on the go. I wish of [ __ ] and then out on the go. I wish they had a better like proper remote they had a better like proper remote they had a better like proper remote control the whole instance feature like control the whole instance feature like control the whole instance feature like codeex does, but it's a quick way to codeex does, but it's a quick way to codeex does, but it's a quick way to just throw the session externally so you just throw the session externally so you just throw the session externally so you can do it from other things. That's kind can do it from other things. That's kind can do it from other things. That's kind of cool. I think that's all I have to of cool. I think that's all I have to of cool. I think that's all I have to say here though. This is a weird say here though. This is a weird say here though. This is a weird reflective experience for me. There's a reflective experience for me. There's a reflective experience for me. There's a lot of things in cloud code that are lot of things in cloud code that are lot of things in cloud code that are good. I went out of my way to not talk good. I went out of my way to not talk good. I went out of my way to not talk about the bad here, but I might have to about the bad here, but I might have to about the bad here, but I might have to in the future. So, if you want me to do in the future. So, if you want me to do in the future. So, if you want me to do a video where I tear apart all of the a video where I tear apart all of the a video where I tear apart all of the things that Cloud Code does wrong, let things that Cloud Code does wrong, let things that Cloud Code does wrong, let me know in the comments. Hope you me know in the comments. Hope you me know in the comments. Hope you enjoyed this one. I need to go crash out enjoyed this one. I need to go crash out enjoyed this one. I need to go crash out about Fable being taken away from me. I about Fable being taken away from me. I about Fable being taken away from me. I am not excited to do that.