WEBVTT NOTE Sentence-level transcript of https://www.youtube.com/watch?v=MSbacZ99E14 NOTE One cue per sentence. Cue ids are the line anchors on /transcripts/MSbacZ99E14.html. A cue ends where the next begins, or 2 s after its last word. s1 00:00:00.240 --> 00:00:03.360 Hi everyone, or should I say hi captains. s2 00:00:03.360 --> 00:00:08.639 Many of you might have already watched my previous videos because they got more than a million views. s3 00:00:08.639 --> 00:00:10.719 If you haven't yet, my name is Kun. s4 00:00:10.719 --> 00:00:15.679 I was an L8 principal engineer previously at Meta, Microsoft, and Atlassian. s5 00:00:15.679 --> 00:00:20.640 My previous videos shared a lot of the basics about agentic engineering. s6 00:00:20.640 --> 00:00:29.039 So this time I'm hoping to share a more advanced workflow focusing on high throughput multitasking through a large number of agents. s7 00:00:29.039 --> 00:00:33.040 I'll walk through it by just showing you how I get some real work done. s8 00:00:33.040 --> 00:00:34.399 Let's jump in. s9 00:00:34.399 --> 00:00:37.040 As always, we will start with a clean terminal. s10 00:00:37.040 --> 00:00:39.520 And the first thing I'm going to launch is Herder. s11 00:00:39.520 --> 00:00:41.680 Herder is basically my new T-Max. s12 00:00:41.680 --> 00:00:42.879 Uh it's a multiplexer. s13 00:00:42.879 --> 00:00:47.440 It manages all my agent sessions and allows me to easily navigate between the sessions. s14 00:00:47.440 --> 00:00:49.600 You will see that in a bit in this video. s15 00:00:49.600 --> 00:00:53.199 And then I'm going to go into my first mate directory. s16 00:00:53.199 --> 00:01:02.000 Uh if you haven't heard of first mate yet uh it is a open source project that I created uh it's got quite some popularity uh a lot of people are using this. s17 00:01:02.000 --> 00:01:04.000 This is basically my workflow. s18 00:01:04.000 --> 00:01:20.080 First mate is the essence of my workflow and uh the core idea is that uh I only talk to one agent which is the first mate and the first mate will orchestrate all the other agents for me so that I don't have to juggle between all the sessions and context switch all the time. s19 00:01:20.080 --> 00:01:23.040 uh is first mate doing that and you will see that in a bit. s20 00:01:23.040 --> 00:01:25.840 Um the way to set up first mate is super easy. s21 00:01:25.840 --> 00:01:29.439 You just just clone this repo and launch your agent in this repo. s22 00:01:29.439 --> 00:01:33.200 Uh so now that I'm in the first mate directory, I just launch my agent. s23 00:01:33.200 --> 00:01:34.560 Uh I use pi. s24 00:01:34.560 --> 00:01:37.439 Um I I use both pi and cloud code. s25 00:01:37.439 --> 00:01:43.439 Uh I basically use cloud code for whenever I want to use an anthropic model uh like opus and fable. s26 00:01:43.439 --> 00:01:48.960 And I use pi for everything else because pi uh basically it can use any other model. s27 00:01:48.960 --> 00:01:52.000 Uh allows me to switch between models very easily as well. s28 00:01:52.000 --> 00:02:05.280 Uh I actually um right now I'm using grock 4.5. Um right now uh as of this day um Grock 4.6 is already available but I actually like 4.5 better because Grock 4.5 s29 00:02:05.280 --> 00:02:08.959 is uh faster and more efficient and more straight to the point. s30 00:02:08.959 --> 00:02:18.640 Uh 4.6 is a little bit like Opus 5. uh it's got uh this weird uh personality where it doesn't quite speak uh like a human. s31 00:02:18.640 --> 00:02:20.959 Um so I like 4.5 better. s32 00:02:20.959 --> 00:02:28.160 Uh and that's something I recommend in general that when the new model comes out, don't assume the new model is always better than the last. s33 00:02:28.160 --> 00:02:36.959 Uh just use it for a bit and get a sense for how the model behaves uh by yourself and uh look at how other people have summarized their learnings as well. s34 00:02:36.959 --> 00:02:43.120 Uh so yeah, this is first mate and this agent session is where I literally do everything. s35 00:02:43.120 --> 00:02:47.040 All my projects, all my tasks, I just use this single agent session to do it. s36 00:02:47.040 --> 00:02:55.599 And maybe to show you what that looks like, I'll just ask first mate list all the projects that we manage through this fleet. s37 00:02:57.200 --> 00:03:01.200 Um yeah, and you notice that I use voice input by default. s38 00:03:01.200 --> 00:03:02.720 Uh and that's how I work. s39 00:03:02.720 --> 00:03:08.560 uh voice input is just faster than uh typing uh most of the time like this kind of natural language prompts. s40 00:03:08.560 --> 00:03:14.080 Uh the only case where I don't use voice is when I have to um type a URL or something like that. s41 00:03:14.080 --> 00:03:14.319 Yeah. s42 00:03:14.319 --> 00:03:19.120 And you can see first mate has already listed all the projects that I currently manage through first mate. s43 00:03:19.120 --> 00:03:20.560 Uh it's a lot of projects. s44 00:03:20.560 --> 00:03:24.720 Uh right uh this is like how many 36 projects in total. s45 00:03:24.720 --> 00:03:28.480 Um so this is uh I do everything uh across these 36 projects. s46 00:03:28.480 --> 00:03:29.760 I do a lot of tasks. s47 00:03:29.760 --> 00:03:33.120 Uh most of these are open source projects that I uh published. s48 00:03:33.120 --> 00:03:35.200 Some of them are private projects as well. s49 00:03:35.200 --> 00:03:37.280 And I do everything just through first mates. s50 00:03:37.280 --> 00:03:41.120 And to show you how I do that, uh maybe I just like walk through some real tasks. s51 00:03:41.120 --> 00:03:46.000 Uh the first thing I probably want to do is u this project called backpass. s52 00:03:46.000 --> 00:03:55.760 Um backpass is a um tool that you can use to help you improve your agent um agents.mmd or cloth.md. s53 00:03:55.760 --> 00:04:06.480 The way this tool works is that it will analyze all your agent sessions in the past and use that to detect what can we improve about your agents.mmd and cloud.md s54 00:04:06.480 --> 00:04:09.120 so that your agents can work better next time. s55 00:04:09.120 --> 00:04:13.680 Um, and this tool uh I'll probably do a dedicated video about this in the future. s56 00:04:13.680 --> 00:04:20.320 Uh, but right now I just want to do one improvement which is uh something I'll just tell first mate. s57 00:04:20.320 --> 00:04:29.440 Hey first mate, previously we planned an idea for backpass which is to allow backpass to collect agent sessions across multiple remote machines. s58 00:04:29.440 --> 00:04:32.000 Um uh I think the plan is already there. s59 00:04:32.000 --> 00:04:37.199 Can we pull that out and then dispatch a crew mate to work on that? s60 00:04:38.639 --> 00:04:39.120 All right. s61 00:04:39.120 --> 00:04:43.120 Uh so that's how I uh ask first mate to do one task. s62 00:04:43.120 --> 00:04:56.800 Uh and first mate will look into um the projects uh and figure out which project this is about and how to uh find the previously um already made plan and use that to delegate to another implement s63 00:04:56.800 --> 00:04:58.240 uh agent to do it. s64 00:04:58.240 --> 00:05:02.160 Um you will notice that um first made agent is already working. s65 00:05:02.160 --> 00:05:04.240 Um but we don't see any tool calls. s66 00:05:04.240 --> 00:05:07.840 We don't see anything in in this uh pi agent terminal, right? s67 00:05:07.840 --> 00:05:18.800 Um and the uh and the only thing we see is this little boat uh that is like floating around and sometimes the agent uh chain of thoughts is showing up as well but it quickly get dismissed. s68 00:05:18.800 --> 00:05:23.199 Um this is done by a custom PI extension that I built. s69 00:05:23.199 --> 00:05:26.560 Uh uh it's uh it comes with a first mate repo. s70 00:05:26.560 --> 00:05:32.720 Um and the reason I do this is that I really don't need to look at what's the agent is doing. s71 00:05:32.720 --> 00:05:34.240 It's making all those tool calls. s72 00:05:34.240 --> 00:05:35.199 It's reading files. s73 00:05:35.199 --> 00:05:36.880 It's doing this and that. s74 00:05:36.880 --> 00:05:42.160 um I don't actually need to know that most of the time those things are distractions and noise. s75 00:05:42.160 --> 00:05:54.080 Uh so I built this uh little custom uh extension called com uh com um and this calm uh command in pi will basically allow me to hide all those noise s76 00:05:54.080 --> 00:05:59.199 and just show this little boat that I uh can allow me to know the agent is working. s77 00:05:59.199 --> 00:06:00.160 It's not stuck. s78 00:06:00.160 --> 00:06:09.520 Um, if I do want to see uh what's kind of tool calls were being made, I can just type uh com uh slashcom to toggle uh this uh details. s79 00:06:09.520 --> 00:06:11.919 So now I can see, oh, it's making this tool call. s80 00:06:11.919 --> 00:06:14.240 It's uh spawning a crew mate, right? s81 00:06:14.240 --> 00:06:16.160 Um so yeah, that's what the agent is doing. s82 00:06:16.160 --> 00:06:23.039 But most of the time I just turn on com mode and I just uh uh don't need to look at all those noise. s83 00:06:23.039 --> 00:06:25.280 Um and then that allows me to focus. s84 00:06:25.280 --> 00:06:30.479 that allows me to actually think about what do I actually need to do next, right? s85 00:06:30.479 --> 00:06:34.400 Uh so that's what's important in not looking at the tool cost and everything. s86 00:06:34.400 --> 00:06:44.000 Um so one thing I do want to uh show here um okay uh first mate said uh it's found uh the plan and dispatched it. s87 00:06:44.000 --> 00:06:52.240 Um so now we can see there is a crew mate here uh right uh in the left hand side panel uh where we can look at all the uh sessions all the agents. s88 00:06:52.240 --> 00:07:02.400 Uh we can see here there is a backpass SSH u this is basically another terminal tab uh where u another agent has spawn up uh and it's running opus 5 s89 00:07:02.400 --> 00:07:12.800 uh it's got the requirements and it's now building the thing uh it's building uh what we uh what I asked this is basically a crew mate uh that's spun up by first mate earlier s90 00:07:12.800 --> 00:07:25.360 um and the reason uh we're using opus 5 here is that uh if we look uh if we toggle com mode off uh and Let's see if we can find the uh previous s91 00:07:25.360 --> 00:07:27.440 uh tool calls. s92 00:07:27.440 --> 00:07:36.080 Um we can probably see that um before first mate decided to um launch the crewmate, it caught a few things. s93 00:07:36.080 --> 00:07:42.400 It called this quotota axi uh and it used a skill called quotota array dispatch. s94 00:07:42.400 --> 00:07:44.560 Uh this is something that's really important. s95 00:07:44.560 --> 00:07:47.199 Uh and uh that's why I want to talk through this. s96 00:07:47.199 --> 00:08:00.400 Um, first mate, uh, when first mate is about to dispatch a crewmate to do something, it will look at a bunch of rules to decide which agent and which LLM to use for that crewmate. s97 00:08:00.400 --> 00:08:06.639 Um, so let me uh turn this back off and uh let me show you and let me go find the config file. s98 00:08:06.639 --> 00:08:11.360 Uh, it's a config file in the config directory called crew dispatch.json. s99 00:08:11.360 --> 00:08:14.400 Um, this is my dis uh dispatch rules. s100 00:08:14.400 --> 00:08:15.840 You can configure yours. s101 00:08:15.840 --> 00:08:19.759 Um, and I actually don't write this rule file by by hand. s102 00:08:19.759 --> 00:08:22.080 I ask first mate to uh do this. s103 00:08:22.080 --> 00:08:27.039 I just tell firstmate what my preferences are and first mate will configure this file for me. s104 00:08:27.039 --> 00:08:35.519 Um, and the rules uh we can look through this for new feature development work on uh this iOS app that I I have always use Fable. s105 00:08:35.519 --> 00:08:39.200 Um, Fable is like really good at building good UI. s106 00:08:39.200 --> 00:08:42.080 Um, and this iOS app actually uh is a paid app. s107 00:08:42.080 --> 00:08:44.720 I really want to make sure it's taken good care of. s108 00:08:44.720 --> 00:08:46.959 Uh so I use Fable for that. s109 00:08:46.959 --> 00:08:51.519 Um and the task requires uh generating images. s110 00:08:51.519 --> 00:08:56.959 Uh if it needs images then I use codeex as the harness and then GPT 5.6 soul. s111 00:08:56.959 --> 00:09:04.240 Um codeex has a native tool for generating images because openai has the uh image generation model right. s112 00:09:04.240 --> 00:09:06.640 Uh so I use codeex to do that. s113 00:09:06.640 --> 00:09:15.519 Um there is also uh when the task is technical product design architecture um or planning work that is like genuinely very difficult. s114 00:09:15.519 --> 00:09:17.360 Uh so it's a difficult planning work. s115 00:09:17.360 --> 00:09:22.160 I use fable I use uh Kim K3 I use Astra. s116 00:09:22.160 --> 00:09:28.160 Uh so I basically allow these three models to do this kind of very complex and ambiguous planning for me. s117 00:09:28.160 --> 00:09:30.000 Uh and there's like Y as well. s118 00:09:30.000 --> 00:09:33.040 Uh so it helps the model uh make the decisions. s119 00:09:33.040 --> 00:09:43.519 uh and then if the task is a simple bug fix uh whose root cause and expected behavior is already well defined then I use Luna I use Sonnet I use uh s120 00:09:43.519 --> 00:09:57.040 cursor uh Gro 4.6 um and then the default if uh a task doesn't fall under any of those rules then the default is uh claude opus and uh this uh cursor and gro 4.6 six high. s121 00:09:57.040 --> 00:10:08.560 Uh so that's basically my uh rules for how I dispatch uh the crewmates and uh when you see I have multiple models defined here right so when uh this rule matches s122 00:10:08.560 --> 00:10:22.640 there is three models that's uh available and when this happens uh the first mate will look at which uh model has the most quotota available and choose the one that has the most quotota. s123 00:10:22.640 --> 00:10:26.480 So I'll show you um how I look at my quota as well. s124 00:10:26.480 --> 00:10:28.399 I use this tool called quota axi. s125 00:10:28.399 --> 00:10:29.920 That's another project I built. s126 00:10:29.920 --> 00:10:36.720 Uh it can give us um this data uh where I can just easily monitor how my quota is doing. s127 00:10:36.720 --> 00:10:38.800 I have a subscription from cloud. s128 00:10:38.800 --> 00:10:39.920 I have a codeex. s129 00:10:39.920 --> 00:10:40.800 I have a cursor. s130 00:10:40.800 --> 00:10:42.560 I have a gro as well. s131 00:10:42.560 --> 00:10:45.519 Uh and this is where they uh sit currently. s132 00:10:45.519 --> 00:10:51.360 Uh I actually took a vacation right before this and I have a ton of kota available uh which is awesome. s133 00:10:51.360 --> 00:11:00.720 Uh so I can build a lot of stuff and this quotota axi is actually um this TUI is for human but uh the default COI is for agent to use. s134 00:11:00.720 --> 00:11:12.160 Uh so first mate will call quota axi to get access to all my quota information and use the model that has the most quota available so that I don't waste any quota. s135 00:11:12.160 --> 00:11:17.440 Uh so this is like really useful because otherwise I have to constantly look at this myself. s136 00:11:17.440 --> 00:11:28.160 Um and with like people like Tibo constantly resetting codeex quota it's really hard to keep track of that right uh so that's why uh I use uh this um way s137 00:11:28.160 --> 00:11:41.600 uh left first mate to manage all this dispatching for me uh and first mate for this case uh it decided to use an opus 5 uh so that's why uh it launches uh this opus 5 crewmates uh under this uh first mate structure s138 00:11:41.600 --> 00:11:49.760 um so I yeah I don't actually look at the the um crewmates um most of the uh I just think about what I want to do next. s139 00:11:49.760 --> 00:11:54.240 Uh there is another project um that I want to uh make some improvements on. s140 00:11:54.240 --> 00:11:56.000 Uh I'll show you. s141 00:11:56.000 --> 00:11:57.680 There is a project I recently built. s142 00:11:57.680 --> 00:12:00.000 I really uh find quite interesting. s143 00:12:00.000 --> 00:12:02.000 Uh it's called Fry with Me. s144 00:12:02.000 --> 00:12:06.959 Uh and this project uh is a uh it's a game basically. s145 00:12:06.959 --> 00:12:08.079 Um and let me show you. s146 00:12:08.079 --> 00:12:09.920 Uh I'll launch a new seed. s147 00:12:09.920 --> 00:12:14.800 Uh it's basically a game, but it's kind of uh it's something to help me relax. s148 00:12:14.800 --> 00:12:22.880 I I I usually put this on my uh second monitor uh when I'm coding and it's just like uh I'll turn off sound. s149 00:12:22.880 --> 00:12:28.639 Uh it's just like a really relaxing uh bird that is flying across the world. s150 00:12:28.639 --> 00:12:31.760 Uh and this whole world is u procedurally generated. s151 00:12:31.760 --> 00:12:33.519 Uh so it's always different. s152 00:12:33.519 --> 00:12:38.959 Uh and I can just put this on a second monitor and like uh keep focusing on my work. s153 00:12:38.959 --> 00:12:46.560 Um, so the thing I want to do with this uh with this game right now uh is that I want to improve some of the visuals uh in this game. s154 00:12:46.560 --> 00:12:54.160 Uh so I'll just go back to um my first mate and say uh tell first mate some of the ideas. s155 00:12:54.160 --> 00:12:59.600 Hey um I want to make some uh improvements to uh fry with me. s156 00:12:59.600 --> 00:13:01.519 Uh I have three ideas. s157 00:13:01.519 --> 00:13:07.440 One is that I want to prototype uh some better looking snowy mountains. s158 00:13:07.440 --> 00:13:12.639 Uh so when there's a mountain with snow uh right now it doesn't look as good. s159 00:13:12.639 --> 00:13:19.519 Uh I want to prototype a few really goodlooking uh mountains um that we can uh render. s160 00:13:19.519 --> 00:13:20.880 Uh so that's one. s161 00:13:20.880 --> 00:13:27.519 Another idea I have is that uh at night right now uh we just show a uh night sky. s162 00:13:27.519 --> 00:13:29.600 We have the Milky Way which is good. s163 00:13:29.600 --> 00:13:32.160 Uh but I also want to have northern lights. s164 00:13:32.160 --> 00:13:35.120 Um they should show up kind of rand randomly. s165 00:13:35.120 --> 00:13:40.079 uh not always uh but uh they should look like uh realistic northern lights. s166 00:13:40.079 --> 00:13:46.480 Um the third idea is that right now there's only one bird and we cannot choose between different kind of birds. s167 00:13:46.480 --> 00:13:51.120 Uh I want to prototype uh a few different kind of birds that people can choose. s168 00:13:51.120 --> 00:13:58.720 Um so yeah for each of these ideas I want you to um prototype three different variants. s169 00:13:58.720 --> 00:14:02.560 Uh so I can choose uh which one uh I like. s170 00:14:02.560 --> 00:14:12.639 Um and uh once the prototypes are ready, uh ask the crew mates to give me uh that in lavish for me to review. s171 00:14:12.639 --> 00:14:18.639 And each idea should be done by a fable crew mate uh in parallel. s172 00:14:20.399 --> 00:14:20.880 All right. s173 00:14:20.880 --> 00:14:24.880 Uh so basically I gave a long ramble of uh ideas that I have. s174 00:14:24.880 --> 00:14:26.720 Uh I like to do this a lot. s175 00:14:26.720 --> 00:14:29.360 Um and you can see I asked for prototypes. s176 00:14:29.360 --> 00:14:36.079 Um and I asked the prototypes to be given to me uh through lavish which we'll look at uh in a bit. s177 00:14:36.079 --> 00:14:42.959 Uh it's a really good way to visually validate uh the ideas and just pick uh very easily and give feedback as well. s178 00:14:42.959 --> 00:14:50.560 Um and I specifically asked for fable uh because uh the reason uh was like uh I was looking at the quota, right? s179 00:14:50.560 --> 00:14:53.040 We have a whole bunch of fable quota available. s180 00:14:53.040 --> 00:14:54.959 Uh so why not uh use that? s181 00:14:54.959 --> 00:15:00.880 Uh so I use fable uh to do this prototypes uh and uh they usually uh can give really good results. s182 00:15:00.880 --> 00:15:01.440 Yeah. s183 00:15:01.440 --> 00:15:09.360 So um this is basically uh while the previous backpass uh task is still ongoing uh the crew mate is still working on that. s184 00:15:09.360 --> 00:15:24.959 uh I just keep giving more and more ideas to first mate and first mate uh is not uh because first mate is not the one that's actually doing the work uh it's always available to talk to me uh and whenever there is some more work uh first mate will just spin up more crew mates to do that s185 00:15:24.959 --> 00:15:39.600 and then come back to me uh and I can talk to it as well um if I have more like more ideas although first mate is working like this boat is still moving uh that's saying like first mate is working I can just keep typing more ideas here I can just talk to first mate type more and s186 00:15:39.600 --> 00:15:41.680 more prompts and kill those prompts. s187 00:15:41.680 --> 00:15:46.800 All right, so I don't actually have to wait for first mate to uh finish up or anything. s188 00:15:46.800 --> 00:15:49.360 Uh I just keep giving more and more prompts. s189 00:15:49.360 --> 00:15:56.399 Um and that's how I uh work with first mate uh to like get a lot of tasks spun up in parallel. s190 00:15:56.399 --> 00:16:07.279 Um something else that I also uh like to do is um when I work through so many projects, there's usually a lot of decisions that were previously discussed but haven't been made. s191 00:16:07.279 --> 00:16:08.000 Uh right. s192 00:16:08.000 --> 00:16:16.639 Um and and sometimes it's really easy to like lose track of which things we talk about that have been done versus have not been done yet. s193 00:16:16.639 --> 00:16:21.360 Um in first mate there is a built-in skill called bearings that can do this. s194 00:16:21.360 --> 00:16:30.800 Uh this will basically allow first mate to collect a snapshot of all the uh active tasks across the fleet and give me a snapshot. s195 00:16:30.800 --> 00:16:35.519 Um but uh the default bearings will give me that uh in the terminal. s196 00:16:35.519 --> 00:16:37.920 Uh which sometimes is like a little bit hard to read. s197 00:16:37.920 --> 00:16:45.199 Uh so I actually have a variant of that called bearings lavish which will get first mate to give me that report in lavish. s198 00:16:45.199 --> 00:16:46.639 Uh so I'll just show you. s199 00:16:46.639 --> 00:16:51.920 Um I used the follow-up prompt to cue this uh right u behind the current actions. s200 00:16:51.920 --> 00:16:54.639 Uh so first mate will not run this yet. s201 00:16:54.639 --> 00:16:56.560 Uh it's queued as a follow-up. s202 00:16:56.560 --> 00:17:03.120 So only when first mate has done the current set of actions then it will uh do the follow-up prompt. s203 00:17:03.120 --> 00:17:08.880 Uh so this is a good way to make sure first mate is not being distracted uh it's actually doing things one by one. s204 00:17:08.880 --> 00:17:14.799 Um if I need to uh actually steer first mate uh to do something I can also steer instead of follow-up. s205 00:17:14.799 --> 00:17:19.520 Uh so followup I think is alt enter and steer is just enter. s206 00:17:19.520 --> 00:17:25.039 So if I type a prompt and enter it will become a steer that will immediately send to first mate. s207 00:17:25.039 --> 00:17:27.679 Um that's a PI feature actually. s208 00:17:27.679 --> 00:17:32.240 Um and uh it's very handy to have a distinction between follow-up and steer. s209 00:17:32.240 --> 00:17:48.000 Um so we can see here first mate has spun up uh backpass this crew mates um and three uh other fly something crew mates right these are the three ideas uh that I asked um just now and first mate is just finishing up uh dispatching all the crew mates s210 00:17:48.000 --> 00:18:01.200 uh and we can check right it's using fable uh so just as we asked u that is something that's very useful is like I can specifically ask for a certain uh agent harness or a certain model to be used s211 00:18:01.200 --> 00:18:04.559 uh for the task task that I uh am about to do. s212 00:18:04.559 --> 00:18:10.480 Um this is useful when I have some opinion or judgment on what is the right model. s213 00:18:10.480 --> 00:18:16.720 Um and um yeah uh for different tasks I can just easily uh get the right model to work on them. s214 00:18:16.720 --> 00:18:22.799 Uh and right right now uh first mate said all those three um fly with me prototypes uh underway. s215 00:18:22.799 --> 00:18:27.200 Uh and now the follow-up prompt bearing stavish uh just uh got sent right. s216 00:18:27.200 --> 00:18:29.840 Uh so that's how a follow-up prompt work. s217 00:18:29.840 --> 00:18:30.559 All right. s218 00:18:30.559 --> 00:18:42.240 First May just brought up this uh lavish uh report um this uh bearings report in lavish um and it's a really uh nicely structured um like uh interactive HTML s219 00:18:42.240 --> 00:18:46.880 that has uh what's uh what's charted next uh what is currently underway. s220 00:18:46.880 --> 00:18:55.440 Uh these are the things that's already happening and uh what's recently landed at the top there's like all these captain's call these are the decisions I have to make. s221 00:18:55.440 --> 00:18:57.840 Uh so let me look through this. s222 00:18:57.840 --> 00:19:02.640 These were usually like things that were previously discussed uh and haven't been done. s223 00:19:02.640 --> 00:19:05.840 So uh it's good to u make a decision on. s224 00:19:05.840 --> 00:19:08.640 So there's a workaround for lavish onboard. s225 00:19:08.640 --> 00:19:12.559 Uh this one yeah I remember this one but I don't want to do that right now. s226 00:19:12.559 --> 00:19:15.760 Uh I just decide claude exit confirm dialogue. s227 00:19:15.760 --> 00:19:17.200 Uh this is for first mate. s228 00:19:17.200 --> 00:19:28.960 Uh and uh Claude recently introduced a new dialogue where um when you try to exit claude when some background tasks are running uh it will show up a dialogue that blocks the exit which is really annoying. s229 00:19:28.960 --> 00:19:34.480 Uh so I think we actually have to do this uh treat as non uh non wedge. s230 00:19:34.480 --> 00:19:36.080 Yep. s231 00:19:36.080 --> 00:19:41.919 Uh this le props up sub uh yeah this one I remember I I don't want to do that right now. s232 00:19:41.919 --> 00:19:45.440 Um uh and I'll just look through like all these decisions. s233 00:19:45.440 --> 00:19:48.240 These were usually like things that we previously discovered. s234 00:19:48.240 --> 00:19:50.400 Uh herder missing end points recreate. s235 00:19:50.400 --> 00:19:51.760 Yep, that's a bug. s236 00:19:51.760 --> 00:19:55.039 Uh keep refuse escalate. s237 00:19:55.039 --> 00:19:59.440 Um yep, that's the right behavior. s238 00:19:59.440 --> 00:20:00.640 Uh URL wrap. s239 00:20:00.640 --> 00:20:02.160 Uh this is for the ship app. s240 00:20:02.160 --> 00:20:06.080 Um and this is a bug uh will uh we'll fix now. s241 00:20:06.080 --> 00:20:07.919 Uh ship a follow-up. s242 00:20:07.919 --> 00:20:16.240 Um this is a another improvement for ship uh for a uh first-time user experience uh to teach user about the multiplexer. s243 00:20:16.240 --> 00:20:17.679 Uh this is quite complex. s244 00:20:17.679 --> 00:20:21.919 Uh so I'm going to keep that held uh and discuss that a little bit deeper. s245 00:20:21.919 --> 00:20:24.240 Uh this is the diaglide banner. s246 00:20:24.240 --> 00:20:26.559 Um yeah, this one I actually want to build now. s247 00:20:26.559 --> 00:20:28.799 Uh so I'll do that. s248 00:20:28.799 --> 00:20:30.559 Uh design kit mock removal. s249 00:20:30.559 --> 00:20:32.960 Yep, I remember this one needs to be done. s250 00:20:32.960 --> 00:20:35.280 Uh I'll decide as well. s251 00:20:35.280 --> 00:20:38.480 Um yeah, you can see there are quite some decisions, right? s252 00:20:38.480 --> 00:20:49.360 Uh because I'm walking through a lot of different projects and uh these projects as I walk through them uh I usually have like a lot of discussions with first mate and some of the decisions I just haven't made. s253 00:20:49.360 --> 00:20:54.159 Um so uh it's good to have this like first mate keeping track of that for me. s254 00:20:54.159 --> 00:20:56.640 Um dial nested fan editor. s255 00:20:56.640 --> 00:20:58.240 Yep, I need that now. s256 00:20:58.240 --> 00:21:00.720 Um what else? s257 00:21:00.720 --> 00:21:02.720 Drag into group. s258 00:21:02.720 --> 00:21:09.679 Uh yeah, this is uh where like um users can customize uh the dial in the ship app. s259 00:21:09.679 --> 00:21:12.080 Um I will do that later. s260 00:21:12.080 --> 00:21:16.960 Um so okay, so all the decisions uh we have recorded uh the follow-up. s261 00:21:16.960 --> 00:21:18.880 Uh so I'll just send this back. s262 00:21:18.880 --> 00:21:20.880 Yeah, we can see here this working spinner. s263 00:21:20.880 --> 00:21:24.320 That means uh the feedback uh was sent back to first mate. s264 00:21:24.320 --> 00:21:27.440 Uh and uh I can actually just end the session now. s265 00:21:27.440 --> 00:21:31.600 Uh if I go back to first mate, yep, first mate is now um having a boat. s266 00:21:31.600 --> 00:21:33.200 uh which means first mate is working. s267 00:21:33.200 --> 00:21:38.640 Um so it's likely that first mate just received the feedback and is now acting on them. s268 00:21:38.640 --> 00:21:42.320 Uh we'll give first mate a bit of time to see uh what first mate does. s269 00:21:42.320 --> 00:21:45.440 Okay, here we can see uh first mate just got back to us. s270 00:21:45.440 --> 00:21:51.600 Uh it said um the uh the four uh decisions about first mate uh is listed here. s271 00:21:51.600 --> 00:21:53.919 Uh actually there's like nothing needed. s272 00:21:53.919 --> 00:21:57.440 Uh the one thing I decided to build was actually already released. s273 00:21:57.440 --> 00:21:59.039 It's already in the codebase. s274 00:21:59.039 --> 00:22:07.200 Um so uh ship uh we decided to um build this now build this now and fix that and prioritize this. s275 00:22:07.200 --> 00:22:15.760 Uh so there's a lot of work on this ship project and we can see um the work is actually being routed to this uh second mate. s276 00:22:15.760 --> 00:22:19.679 We can see there's a crew mate under here that's uh currently running. s277 00:22:19.679 --> 00:22:24.400 Um second mate is basically a way to scale up uh first mate's charter. s278 00:22:24.400 --> 00:22:32.720 Um so you can see as we give more and more tasks to first mate uh we already have four tasks right um if we keep giving more and more tasks s279 00:22:32.720 --> 00:22:48.960 um to first mate at some point you will realize oh first mate is like way too busy uh because it's always orchestrating those tasks and have no time to even talk to me um so at some point you will have to scale up and this is uh what I did I basically created a whole bunch of second s280 00:22:48.960 --> 00:22:57.840 mates to own different domains um of the charter so these second mates uh ship uh owns everything related to this ship app. s281 00:22:57.840 --> 00:23:02.640 So every uh piece of work that's on this ship app will get routed to this second mate. s282 00:23:02.640 --> 00:23:05.120 So first mate doesn't have to orchestrate all that. s283 00:23:05.120 --> 00:23:09.840 Uh it can just send the ask to this second mate and the second mate will do it. s284 00:23:09.840 --> 00:23:17.360 Um and we can see here first this second mate is already has already spun up this crew mate uh for the URL rep bug fix. s285 00:23:17.360 --> 00:23:21.360 Uh there's a few more that will likely be spun up soon as well. s286 00:23:21.360 --> 00:23:25.360 Um, every second mate is actually a first mate by itself. s287 00:23:25.360 --> 00:23:29.760 Um, so I can actually just go into this second mate and see what it's doing. s288 00:23:29.760 --> 00:23:32.880 Uh, I can talk to it just like a normal first mate. s289 00:23:32.880 --> 00:23:35.840 Um, and uh, maybe uh, let me show one example. s290 00:23:35.840 --> 00:23:39.440 Uh, this is uh, a second mate for the Edis wallets app. s291 00:23:39.440 --> 00:23:42.720 Uh, this app was the app I built in my last video. s292 00:23:42.720 --> 00:23:45.120 This app is actually already released in the app store. s293 00:23:45.120 --> 00:23:49.280 U, but I actually forgot uh, which features got in versus not. s294 00:23:49.280 --> 00:24:00.000 So, I'll just ask uh which features have already landed that's not released to the app store yet, right? s295 00:24:00.000 --> 00:24:02.320 Um so I can just talk to this second mate. s296 00:24:02.320 --> 00:24:04.640 Every second mate is also a first mate. s297 00:24:04.640 --> 00:24:06.960 Uh so it will manage its own crew mates. s298 00:24:06.960 --> 00:24:10.559 Uh it has its own memory and uh instructions and everything. s299 00:24:10.559 --> 00:24:12.480 Um so I can just talk to it. s300 00:24:12.480 --> 00:24:16.000 Uh I can also talk to it through first mate, right? s301 00:24:16.000 --> 00:24:23.440 I can ask first mates the same question here and first mate will route the question to that second mate uh whenever appropriate. s302 00:24:23.440 --> 00:24:28.000 Most of the time uh I actually go with the easy route where I just talk to first mate. s303 00:24:28.000 --> 00:24:28.240 Yeah. s304 00:24:28.240 --> 00:24:33.760 So this second mate is now uh trying to answer our question uh and we can let it do its work. s305 00:24:33.760 --> 00:24:38.480 Um and you can see here um there is a local versus mini. s306 00:24:38.480 --> 00:24:40.400 Uh this is a herder feature. s307 00:24:40.400 --> 00:24:46.240 So Herder has the ability to manage all the agent sessions across multiple machines. s308 00:24:46.240 --> 00:24:52.159 Um so this mini is my Mac Mini uh which is uh uh actually on my shelf right now. s309 00:24:52.159 --> 00:24:54.080 Um it's running headlessly. s310 00:24:54.080 --> 00:24:58.159 Uh so I don't connect any monitors or mouse or keyboard to the Mac Mini. s311 00:24:58.159 --> 00:25:00.640 I just let it run uh headlessly. s312 00:25:00.640 --> 00:25:05.760 And this mini actually manages a lot of uh have a lot of second mates running on it. s313 00:25:05.760 --> 00:25:14.159 Uh so um the mini has the second mate for first mate development for ship for Eddie's wallet for axi for uh there's a default one as well. s314 00:25:14.159 --> 00:25:20.080 Um so the mini uh has uh basically more hardware resources than my MacBook. s315 00:25:20.080 --> 00:25:22.240 This current local machine is a MacBook. s316 00:25:22.240 --> 00:25:26.400 Uh it has uh a little bit less power uh than what my mini has. s317 00:25:26.400 --> 00:25:32.320 Uh so I assigned a lot of domains uh to the Mac Mini uh so it can do more work. s318 00:25:32.320 --> 00:25:32.640 Okay. s319 00:25:32.640 --> 00:25:35.760 Uh there's actually a lavish board that's just brought up to me. s320 00:25:35.760 --> 00:25:39.279 Um and we can see it's the bird kinds for fly with me. s321 00:25:39.279 --> 00:25:41.600 Uh this is what I asked earlier, right? s322 00:25:41.600 --> 00:25:43.039 Uh and it's done by Fable. s323 00:25:43.039 --> 00:25:45.039 Uh so now it's ready for review. s324 00:25:45.039 --> 00:25:48.640 Um it's really nice to have this lavish uh interactive artifact. s325 00:25:48.640 --> 00:25:51.039 You will see um the current bird, right? s326 00:25:51.039 --> 00:25:52.240 This is the current bird. s327 00:25:52.240 --> 00:25:56.559 It took screenshots so that I can see what's the current bird looks like. s328 00:25:56.559 --> 00:26:00.480 Uh and board soaring raptor. s329 00:26:00.480 --> 00:26:01.840 Okay, this is the prototype, right? s330 00:26:01.840 --> 00:26:03.679 So, this is a new kind of bird. s331 00:26:03.679 --> 00:26:05.840 Uh, and it's giving me this preview. s332 00:26:05.840 --> 00:26:07.200 Uh, this is really nice. s333 00:26:07.200 --> 00:26:11.279 Uh, I I really like um looking at this previews. s334 00:26:11.279 --> 00:26:13.039 Um, so I actually like this one. s335 00:26:13.039 --> 00:26:14.880 Uh, this this is okay. s336 00:26:14.880 --> 00:26:18.159 Uh, what is this slender swallow? s337 00:26:18.159 --> 00:26:20.799 Uh, what is a what is a slender swallow? s338 00:26:20.799 --> 00:26:23.200 I I don't know either of those words. s339 00:26:23.200 --> 00:26:26.400 Um, but uh it it looks nice. s340 00:26:26.400 --> 00:26:27.440 Uh, it's okay. s341 00:26:27.440 --> 00:26:29.840 This one I probably will also keep. s342 00:26:29.840 --> 00:26:32.799 Uh there is a large ocean glider. s343 00:26:32.799 --> 00:26:35.120 Okay, this is just like the current one, right? s344 00:26:35.120 --> 00:26:38.799 Like it's uh a little bit different but very similar. s345 00:26:38.799 --> 00:26:42.240 Okay, there's a long necked crane. s346 00:26:42.240 --> 00:26:43.200 Oh, cranes. s347 00:26:43.200 --> 00:26:44.559 I like cranes. s348 00:26:44.559 --> 00:26:46.159 Uh they are good. s349 00:26:46.159 --> 00:26:48.880 What is this soft night owl? s350 00:26:48.880 --> 00:26:50.640 Wow, it has so many prototypes. s351 00:26:50.640 --> 00:26:52.480 It's done a lot of work for us. s352 00:26:52.480 --> 00:26:53.360 Um so, okay. s353 00:26:53.360 --> 00:26:54.799 Uh so there's an owl. s354 00:26:54.799 --> 00:26:56.400 I kind of like this owl as well. s355 00:26:56.400 --> 00:26:58.240 It's like round and cute. s356 00:26:58.240 --> 00:27:02.159 Um, so, uh, what do we need to look at here? s357 00:27:02.159 --> 00:27:03.440 Each kind closer. s358 00:27:03.440 --> 00:27:04.960 Okay, closer. s359 00:27:04.960 --> 00:27:07.440 Uh, okay, that's closer. s360 00:27:07.440 --> 00:27:12.799 Um, it actually explained, uh, like the key features, um, of each, um, bird. s361 00:27:12.799 --> 00:27:14.159 Uh, that's really nice. s362 00:27:14.159 --> 00:27:19.440 Uh, but I think I already like have seen enough about these birds to know what I like. s363 00:27:19.440 --> 00:27:23.440 Uh, usually at the bottom it will list uh, the decisions for me to make, right? s364 00:27:23.440 --> 00:27:24.559 So, recommendations. s365 00:27:24.559 --> 00:27:26.720 Uh, carry forward, carry forward. s366 00:27:26.720 --> 00:27:43.600 Uh oh, it actually um it actually recommended to iterate before deciding um okay so uh let's pick right um current bird uh current birds we don't need to do anything right um s367 00:27:43.600 --> 00:27:57.440 broad this raptor yes uh the swallow I uh I I like it um the glider uh the glider is kind of like uh the same as the current one so I'm not going uh want it. s368 00:27:57.440 --> 00:27:58.720 Uh I need the crane. s369 00:27:58.720 --> 00:27:59.760 I need the owl. s370 00:27:59.760 --> 00:28:00.399 Okay. s371 00:28:00.399 --> 00:28:03.120 Uh so yeah, these are our decisions. s372 00:28:03.120 --> 00:28:06.480 And we'll just kill this uh and send it back. s373 00:28:06.480 --> 00:28:12.640 Uh I'll just say like uh let's build uh the birds we decided. s374 00:28:12.640 --> 00:28:13.520 Okay. s375 00:28:13.520 --> 00:28:14.799 And send it an end. s376 00:28:14.799 --> 00:28:15.600 And that's it. s377 00:28:15.600 --> 00:28:17.520 That's my feedback back to Fable. s378 00:28:17.520 --> 00:28:26.480 and Fable um that crewmate who uh prototyped this should uh receive my feedback about which birds that we wanted to forward. s379 00:28:26.480 --> 00:28:28.399 Um and uh that's it. s380 00:28:28.399 --> 00:28:33.200 Uh so that's the um only thing we had to do with that prototype. s381 00:28:33.200 --> 00:28:38.080 Um we can see here uh ship has already spun up three ship crew mates, right? s382 00:28:38.080 --> 00:28:40.080 Uh working on three things in parallel. s383 00:28:40.080 --> 00:28:41.760 So right now how many do we have? s384 00:28:41.760 --> 00:28:45.039 We have one, two, three, four, uh five, six, seven. s385 00:28:45.039 --> 00:28:47.679 We have seven crew mates working in parallel. s386 00:28:47.679 --> 00:28:49.600 Um, and we can keep giving more. s387 00:28:49.600 --> 00:28:51.600 Like first mate is idle right now, right? s388 00:28:51.600 --> 00:28:52.880 First mate is not doing anything. s389 00:28:52.880 --> 00:28:54.720 We can just keep giving more ideas. s390 00:28:54.720 --> 00:28:57.120 Um, I uh that's what I will typically do. s391 00:28:57.120 --> 00:28:59.039 I just keep giving more and more. s392 00:28:59.039 --> 00:29:03.120 Um, and this edi wallet just answered our question as well. s393 00:29:03.120 --> 00:29:07.120 The second mate, uh, it landed since this not released. s394 00:29:07.120 --> 00:29:12.480 Uh, there's a few Okay, there's a few new userfacing features that's landed but not released. s395 00:29:12.480 --> 00:29:14.080 Uh, auto pay local. s396 00:29:14.080 --> 00:29:15.760 Oh, these are not released yet. s397 00:29:15.760 --> 00:29:16.480 Okay. s398 00:29:16.480 --> 00:29:17.279 Uh yeah. s399 00:29:17.279 --> 00:29:20.960 So now I can decide uh do I want to like release those features. s400 00:29:20.960 --> 00:29:23.520 Uh but I'll probably put that aside for now. s401 00:29:23.520 --> 00:29:26.480 Uh we have enough uh to like work through here. s402 00:29:26.480 --> 00:29:26.720 Yeah. s403 00:29:26.720 --> 00:29:39.440 So while we wait for these current uh work items to uh finish, uh something I want to show you that's really cool uh is that uh I have a Discord channel here uh built with We actually have a lot of cool people uh in this server now. s404 00:29:39.440 --> 00:29:46.159 Uh there's a lot of people online and uh I think there's like more than a thousand 2,000 people now uh in the server. s405 00:29:46.159 --> 00:29:53.440 Um so um we have um like constant discussions about first made about uh models and harnesses uh as well. s406 00:29:53.440 --> 00:30:03.919 Uh it's a really cool community um and if you uh want to build with agents uh you enjoy discussing like tips and tricks with others uh this discord server is a really good place. s407 00:30:03.919 --> 00:30:12.399 Uh but the thing I want to show you is like uh I have this bug reports channel uh that I usually come here uh to look at what things uh I should be fixing. s408 00:30:12.399 --> 00:30:15.760 Um let me see uh what I haven't looked at yet. s409 00:30:15.760 --> 00:30:19.279 Um I think this one I um probably haven't looked at yet. s410 00:30:19.279 --> 00:30:21.600 So this one uh what is this? s411 00:30:21.600 --> 00:30:23.600 Um uh it's a treehouse. s412 00:30:23.600 --> 00:30:23.840 Okay. s413 00:30:23.840 --> 00:30:25.840 Treehouse is another project I have. s414 00:30:25.840 --> 00:30:30.240 Uh and uh it's a bug report saying this uh hook is not firing. s415 00:30:30.240 --> 00:30:31.679 Okay. s416 00:30:31.679 --> 00:30:35.360 So uh it seems there's a lot of debug information already available. s417 00:30:35.360 --> 00:30:39.840 So something I will show you here is that I can actually add my first mate. s418 00:30:39.840 --> 00:30:41.840 Uh so I can just uh there's a bot here. s419 00:30:41.840 --> 00:30:43.200 Uh that's my first mate. s420 00:30:43.200 --> 00:30:48.880 I can just ask my first mate uh to uh investigate this, right? s421 00:30:48.880 --> 00:30:51.600 Uh so I can just uh add my first mate here. s422 00:30:51.600 --> 00:31:01.279 And this bot is actually relay uh relaying this uh message back to my actual first mate here running this local session. s423 00:31:01.279 --> 00:31:06.080 Uh so this uh boat that's moving is likely uh from that mention. s424 00:31:06.080 --> 00:31:14.080 So u I can just like basically summon my first mate uh this single agent session that I have from different places. s425 00:31:14.080 --> 00:31:18.640 Uh I can right now it supports discord and uh x as well. s426 00:31:18.640 --> 00:31:20.399 Uh maybe I can also show you that. s427 00:31:20.399 --> 00:31:33.600 Uh so on X uh I also sometimes uh when people report problems uh right so this uh uh a person reporting something uh saying confused by settings send feedback uh in this ship app s428 00:31:33.600 --> 00:31:49.760 uh okay so here I can also just uh add my first mate uh right and say uh let's improve uh this confusing uh message. s429 00:31:52.080 --> 00:31:53.279 All right. s430 00:31:53.279 --> 00:31:54.480 Um, yeah. s431 00:31:54.480 --> 00:32:00.159 So, this at uh this mention will also get relayed to my first mate session. s432 00:32:00.159 --> 00:32:00.399 Yeah. s433 00:32:00.399 --> 00:32:06.320 So, now if we come back to this Discord thread, uh we should see first mate reply uh very soon. s434 00:32:06.320 --> 00:32:12.799 Now, yeah, it just replied uh I captain looking into the treehouse predestroy hook uh not firing on destroy. s435 00:32:12.799 --> 00:32:14.000 Uh we'll report back. s436 00:32:14.000 --> 00:32:14.240 Yeah. s437 00:32:14.240 --> 00:32:20.880 So now if we come back to this X thread, uh first mate also replied uh saying I captain clarifying that send feedback version message. s438 00:32:20.880 --> 00:32:21.360 Uh right. s439 00:32:21.360 --> 00:32:22.399 So uh that's really good. s440 00:32:22.399 --> 00:32:26.880 Uh this is just like how I uh work across different surfaces. s441 00:32:26.880 --> 00:32:34.080 It's so nice to have all these messages from all these different channels and surfaces all funneled into this single agent session. s442 00:32:34.080 --> 00:32:34.480 Right? s443 00:32:34.480 --> 00:32:41.600 So this first made agent session literally holds all my intent and all the context across every project that we have. s444 00:32:41.600 --> 00:32:49.279 Uh so it can play a really good role in coordinating across everything and orchestrating all the crew uh to get things done for us. s445 00:32:49.279 --> 00:32:57.760 Okay, while we wait for um the other crew mates to come back to us, uh maybe one thing that I want to uh quickly talk through here is the context window. s446 00:32:57.760 --> 00:33:04.480 Um so here you can see um the Gro 4.5 uh the default context window limits is 500k tokens. s447 00:33:04.480 --> 00:33:07.600 So here it says we already used more than 40% of that. s448 00:33:07.600 --> 00:33:22.080 Um and a lot of people ask me um whether uh they should u manually manage this context window and when they should compact when they should reset uh and everything to keep uh the context window uh efficient and optimized. s449 00:33:22.080 --> 00:33:31.760 Um so my take on this is that there's a lot of things you can do uh to manage this context window and optimize that and get some savings. s450 00:33:31.760 --> 00:33:38.399 Um, sometimes you can get the agent to use less context to do the same things uh and get things done with less cost. s451 00:33:38.399 --> 00:33:50.000 Uh, sometimes you can also um uh optimize this uh so that the agent is a little bit smarter because uh if you have a lot of context loaded, it can degrade your agent's performance as well. s452 00:33:50.000 --> 00:33:51.840 Um, so there's a lot of things you can do. s453 00:33:51.840 --> 00:33:58.000 Um, but my take is that I actually advise against uh managing that yourself manually. s454 00:33:58.000 --> 00:34:07.360 The reason is that our most precious resource right now, the biggest bottleneck across everything we do with agents is our attention and time. s455 00:34:07.360 --> 00:34:09.679 Uh our human attention and time. s456 00:34:09.679 --> 00:34:21.839 So uh if you use your mental bandwidth to deal with this like little optimizations here and there, then you don't have enough time and enough uh energy to think about what should we build s457 00:34:21.839 --> 00:34:24.320 and that is actually the most important thing. s458 00:34:24.320 --> 00:34:27.520 Um so I actually think you can just waste a little bit. s459 00:34:27.520 --> 00:34:31.359 It's okay to not be optimal about everything, right? s460 00:34:31.359 --> 00:34:37.280 Um there's um there's going to be tools, there's going to be things that will help us manage that automatically. s461 00:34:37.280 --> 00:34:45.280 Um and uh eventually things will just catch up and those features will get built into the harness and we don't have to worry about that anymore. s462 00:34:45.280 --> 00:34:51.040 So right now uh I actually suggest what you do is that you set a constant threshold. s463 00:34:51.040 --> 00:34:55.200 Um so 500k for grock is actually a very reasonable threshold. s464 00:34:55.200 --> 00:35:04.880 Um for codeex uh for cloud uh cloud code may be a little bit different uh because cloud code uh has a default of 1 million tokens as the context window limit. s465 00:35:04.880 --> 00:35:07.200 That is a little bit too high I would say. s466 00:35:07.200 --> 00:35:10.000 Um so uh cloud code uh there's different settings. s467 00:35:10.000 --> 00:35:15.440 I think there's a environment variable you can set to uh limit your context window size. s468 00:35:15.440 --> 00:35:17.520 Uh let me see if I can find it actually. s469 00:35:17.520 --> 00:35:20.480 Um I have that in my dot files. s470 00:35:20.480 --> 00:35:34.400 uh files uh if I search for auto compaction uh yeah so this is the environment variable you can set cloud code autoco compact window uh and I have that set to 500k s471 00:35:34.400 --> 00:35:48.079 um so that means every time the session reaches 500k tokens it will automatically compact um and that's all I suggest right now um you just set a threshold that allows the agent to automatically compact s472 00:35:48.079 --> 00:35:53.040 uh the session so it doesn't become too expensive and too uh inefficient. s473 00:35:53.040 --> 00:35:55.200 Um otherwise just forget about it. s474 00:35:55.200 --> 00:35:57.680 Don't think about the context window too much. s475 00:35:57.680 --> 00:36:03.599 If your brain is thinking about the context window all the time, you don't have enough time to think about what to build. s476 00:36:03.599 --> 00:36:12.480 Um so I want you to optimize your time and attention on what is the right thing to build because that is the most important thing. s477 00:36:12.480 --> 00:36:19.839 If you do that right, you will get back much much more value than what you can save with this little tricks. s478 00:36:19.839 --> 00:36:21.520 Uh I hope that makes sense. s479 00:36:21.520 --> 00:36:26.079 Oh, and here uh actually first mate reported uh the treehouse finding right. s480 00:36:26.079 --> 00:36:28.240 So uh treehouse investigation is done. s481 00:36:28.240 --> 00:36:30.400 Finding is here not a destroy bug. s482 00:36:30.400 --> 00:36:34.800 Uh real defect is hooks uh intentionally discarded for safety. s483 00:36:34.800 --> 00:36:35.599 What? s484 00:36:35.599 --> 00:36:38.079 Uh and that drop is silent. s485 00:36:38.079 --> 00:36:39.599 We dropped hooks. s486 00:36:39.599 --> 00:36:42.240 Uh okay. s487 00:36:42.240 --> 00:36:46.000 Oh, so hooks in the repo is being discarded. s488 00:36:46.000 --> 00:36:48.960 Uh destroy never run. s489 00:36:48.960 --> 00:36:50.400 Keep ignoring repo hooks. s490 00:36:50.400 --> 00:36:53.440 Uh I now remembered that is a security thing. s491 00:36:53.440 --> 00:36:54.079 Yes. s492 00:36:54.079 --> 00:36:57.440 Um so emit a one-time warning. s493 00:36:57.440 --> 00:36:59.680 Um okay. s494 00:36:59.680 --> 00:37:01.760 Uh yes. s495 00:37:01.760 --> 00:37:09.839 Uh let's ship the warning fix and also reply uh to that Discord thread uh for what is happening here. s496 00:37:10.560 --> 00:37:17.839 Um yeah so uh we'll do that uh and that is the closure for the treehouse uh buck report. s497 00:37:17.839 --> 00:37:22.560 Um so there there's a lot of bug reports coming in discord uh in GitHub issues. s498 00:37:22.560 --> 00:37:30.800 Uh I just funnel them all into my agent session uh here in the first mates and that first mate helped me do the investigations uh and all the stuff. s499 00:37:30.800 --> 00:37:37.200 Uh I didn't even have to look into the crew mates that did the investigation here right uh so that is the benefit of first mate. s500 00:37:37.200 --> 00:37:41.280 I just talk to first mate and first mate will orchestrate everything for me. s501 00:37:41.280 --> 00:37:44.800 Uh I don't have to like actually juggle between all these sessions. s502 00:37:44.800 --> 00:37:52.000 uh how many sessions like we had like uh I think at some point we had nine sessions or 10 sessions and most of the time I don't really look at any of them. s503 00:37:52.000 --> 00:37:53.839 Oh, first mate gave another update. s504 00:37:53.839 --> 00:37:58.240 Uh so um there is this uh treehouse warning fix is dispatched. s505 00:37:58.240 --> 00:38:00.240 Uh discord update is posted. s506 00:38:00.240 --> 00:38:02.720 Um fry with me bird clients merged. s507 00:38:02.720 --> 00:38:03.760 Okay. s508 00:38:03.760 --> 00:38:05.440 Uh if you missed it, I did miss it. s509 00:38:05.440 --> 00:38:11.040 Uh so there's a PR and we can see uh this PR didn't even ask for my review. s510 00:38:11.040 --> 00:38:12.480 Uh and I'll show you why. s511 00:38:12.480 --> 00:38:15.920 um is because uh I told first mate to do do that. s512 00:38:15.920 --> 00:38:17.760 Uh I'll show you the rules. s513 00:38:17.760 --> 00:38:31.040 Uh so in first mate uh I think it's in config uh config or in data in data um projects uh MD. s514 00:38:31.040 --> 00:38:36.240 Uh so in this project MD there is a policy being set on every single project. s515 00:38:36.240 --> 00:38:44.800 Uh so if I search for fly with me uh this is uh policy that I set is called direct PR plus yolo. s516 00:38:44.800 --> 00:38:57.839 Um so this basically says first mate you make your judgment calls on this uh and every change should raise a PR um without going through no mistakes uh and just yolo merge if it looks good. s517 00:38:57.839 --> 00:39:08.240 Um so this is for projects where uh I don't think uh an a problem will actually cause any harm uh because this fly with me uh project is just like a toy s518 00:39:08.240 --> 00:39:10.160 u no one is actually depending on it. s519 00:39:10.160 --> 00:39:21.599 So yeah so you can see every project has a different policy here uh for most of the project I have uh I have no mistakes on uh and uh some of them have yolo some of them don't have yolo. s520 00:39:21.599 --> 00:39:28.480 The ones that have YOLO on is usually the projects where a code change will not get released immediately. s521 00:39:28.480 --> 00:39:32.240 Um and we have a chance to test them out before we uh release them. s522 00:39:32.240 --> 00:39:44.320 Uh so yeah, this is where um the policies are being managed and you can tell your first mate as well which projects actually need no mistakes, which projects actually uh need your eyes before merging a PR. s523 00:39:44.320 --> 00:39:50.240 Um so this is very um helpful for managing different levels of uh guardrails. s524 00:39:50.240 --> 00:39:52.240 Uh not everything need no mistakes. s525 00:39:52.240 --> 00:39:54.800 Not everything need uh a uh human review. s526 00:39:54.800 --> 00:39:57.440 Uh depending on how high stake that is. s527 00:39:57.440 --> 00:40:12.240 Uh and for no mistakes uh for those of you who are already familiar with that no mistakes uh is a very strong validation pipeline that will do adversarial review and uh make sure the code change is really safe to merge. s528 00:40:12.240 --> 00:40:15.200 Uh that takes a long time and use a lot of tokens. s529 00:40:15.200 --> 00:40:19.280 So I don't actually always use no mistakes uh for every change I do. s530 00:40:19.280 --> 00:40:26.480 A good way to think about this is like for this code change, would you have asked a another human to review this code? s531 00:40:26.480 --> 00:40:31.760 Uh if you would not have have another code reviewer, then you probably don't need no mistakes. s532 00:40:31.760 --> 00:40:43.839 Um if the code change is risky and high stake enough that you would otherwise ask a human peer to review, then it's probably worth uh activating no mistakes to make sure that's really good. s533 00:40:43.839 --> 00:40:50.480 Um, and I'll probably like show you an example of a PR that's merged through no mistakes for you to see what that looks like. s534 00:40:50.480 --> 00:40:54.160 Um, there's actually another uh ship uh change that's landed as well. s535 00:40:54.160 --> 00:40:56.319 Uh, the design kit mock removal. s536 00:40:56.319 --> 00:40:57.359 Uh, good. s537 00:40:57.359 --> 00:40:58.400 That's good. s538 00:40:58.400 --> 00:41:04.000 Um, I think we just had another uh lavish artifact ready for review as well. s539 00:41:04.000 --> 00:41:06.800 This is the snow capped mountains uh that I asked for. s540 00:41:06.800 --> 00:41:10.160 Uh, I'm very excited to look at what this looks like. s541 00:41:10.160 --> 00:41:12.000 Uh, what the snow does today. s542 00:41:12.000 --> 00:41:12.160 Okay. s543 00:41:12.160 --> 00:41:14.000 Yeah, I don't need to know that. s544 00:41:14.000 --> 00:41:22.000 Uh yeah, today uh it's good to uh for you to see that it's uh like this and it's not like very mountain uh looking, right? s545 00:41:22.000 --> 00:41:26.000 That's the reason I asked for a better snowy mountain. s546 00:41:26.000 --> 00:41:30.319 Uh Fable Fable is like really uh careful here. s547 00:41:30.319 --> 00:41:33.280 It's like done some calculations and everything. s548 00:41:33.280 --> 00:41:37.280 Uh and it's it's got a good like climate model. s549 00:41:37.280 --> 00:41:39.599 Uh interesting. s550 00:41:39.599 --> 00:41:41.599 um three shading studies. s551 00:41:41.599 --> 00:41:42.880 I did a lot of studies. s552 00:41:42.880 --> 00:41:44.640 That's why it took so much time. s553 00:41:44.640 --> 00:41:47.359 So there is this like painted alpine. s554 00:41:47.359 --> 00:41:48.079 Okay. s555 00:41:48.079 --> 00:41:50.560 There's this uh story book cap. s556 00:41:50.560 --> 00:41:51.520 Okay. s557 00:41:51.520 --> 00:41:52.720 Ah interesting. s558 00:41:52.720 --> 00:41:57.920 Uh uh there's this glacial jade. s559 00:41:57.920 --> 00:42:00.800 Uh how's this different from the other ones? s560 00:42:00.800 --> 00:42:02.880 It looks like kind of similar. s561 00:42:02.880 --> 00:42:05.839 Uh and uh what's this? s562 00:42:05.839 --> 00:42:06.960 Carved peaks. s563 00:42:06.960 --> 00:42:08.160 The terrain study. s564 00:42:08.160 --> 00:42:09.119 Oh. s565 00:42:09.119 --> 00:42:11.119 Okay. s566 00:42:11.200 --> 00:42:14.400 Uh oh, this looks good. s567 00:42:14.400 --> 00:42:16.960 Uh this is uh like it'll go above the cloud. s568 00:42:16.960 --> 00:42:20.000 Uh that is also nice. s569 00:42:20.079 --> 00:42:21.359 Uh compare. s570 00:42:21.359 --> 00:42:26.960 We can actually um preview these different uh variants. s571 00:42:29.839 --> 00:42:30.880 This is so nice. s572 00:42:30.880 --> 00:42:39.359 like um uh it's just really good to have this visual uh validation for what this is actually going to look like. s573 00:42:39.359 --> 00:42:42.640 Uh it's a really good way to discuss prototypes. s574 00:42:42.640 --> 00:42:45.680 Um I actually quite like uh these curved peaks. s575 00:42:45.680 --> 00:42:47.920 Uh this uh look pretty nice. s576 00:42:47.920 --> 00:42:52.319 Um but I don't uh want to accept this just yet. s577 00:42:52.319 --> 00:42:59.200 Um I think I actually want this uh the mountain peaks to look more like Mount Everest. s578 00:42:59.200 --> 00:43:01.440 that kind of mountain if you know what I mean. s579 00:43:01.440 --> 00:43:06.319 Um so I will uh give some feedback here. s580 00:43:06.480 --> 00:43:08.079 Uh this is mostly good. s581 00:43:08.079 --> 00:43:09.280 This is a good improvements. s582 00:43:09.280 --> 00:43:14.079 Uh but I actually want the mountain peaks to look more like Mount Everest. s583 00:43:14.079 --> 00:43:27.520 Uh can you go like find some uh photos of Mount Everest uh and K2 um those mountain peaks and try to get our uh mountains to be closer to that looking s584 00:43:28.800 --> 00:43:29.359 All right. s585 00:43:29.359 --> 00:43:31.920 Uh, so that feedback goes back to the fable crew mates. s586 00:43:31.920 --> 00:43:33.680 Uh, so that's my feedback. s587 00:43:33.680 --> 00:43:40.319 Um, and let's, uh, come back here and we can see, uh, treehouse warning fix, uh, is also underway. s588 00:43:40.319 --> 00:43:46.720 Maybe one thing I'll show you is like, uh, you can see like we went to the lavish artifact to review stuff, right? s589 00:43:46.720 --> 00:43:52.079 Uh, and while we are reviewing, uh, the first mate has already like printed a lot of information. s590 00:43:52.079 --> 00:43:52.560 Uh, right? s591 00:43:52.560 --> 00:43:54.319 It's got so many things. s592 00:43:54.319 --> 00:43:58.240 uh and across this whole session, did I miss something, right? s593 00:43:58.240 --> 00:44:01.119 Did I actually look through everything First Mate told me? s594 00:44:01.119 --> 00:44:02.480 It's really unclear. s595 00:44:02.480 --> 00:44:05.920 And this is where the chat UX is a little bit limiting. s596 00:44:05.920 --> 00:44:12.240 Um so what I have here is like in First Mate repo, there is a built-in skill called AOY. s597 00:44:12.240 --> 00:44:15.200 It's just a really good way to catch up on things. s598 00:44:15.200 --> 00:44:22.560 Um so a Hoy basically tells First Mate to summarize what First Mate has told you since your last message. s599 00:44:22.560 --> 00:44:26.880 and uh also pull out any open decisions that you haven't made yet. s600 00:44:26.880 --> 00:44:27.119 Right? s601 00:44:27.119 --> 00:44:36.400 So uh here we can see uh it uh told us uh it's landed uh this bird kinds uh this uh mock removal uh there's a bunch of things still underway s602 00:44:36.400 --> 00:44:38.720 uh and there's still some things waiting for me. s603 00:44:38.720 --> 00:44:43.280 Uh and these things uh I actually directly gave feedback uh to the crew mate. s604 00:44:43.280 --> 00:44:45.040 So uh it should be okay. s605 00:44:45.040 --> 00:44:47.920 Uh there's an open call uh snowy mountains. s606 00:44:47.920 --> 00:44:48.160 Yep. s607 00:44:48.160 --> 00:45:00.720 Uh so um the uh the reason uh first mate doesn't know we already made a decision on snowy mountains uh I already gave feedback uh to uh the crew mates is because I gave that directly through lavish s608 00:45:00.720 --> 00:45:13.520 and that goes straight to the crew mates and doesn't uh go through first mate uh so first mate thinks uh it's still open uh which is okay um and uh very soon the crewmate will give an update and first mate will catch up as well. s609 00:45:13.520 --> 00:45:19.040 Uh so this is basically a good way for uh me to get a sense uh did I miss something? s610 00:45:19.040 --> 00:45:20.319 Is everything okay? s611 00:45:20.319 --> 00:45:23.760 Uh just a hoy and I can just keep spamming a hoy. s612 00:45:23.760 --> 00:45:25.359 It's a very cheap uh call. s613 00:45:25.359 --> 00:45:27.760 It's this skill doesn't cost too many tokens. s614 00:45:27.760 --> 00:45:31.119 Uh it's basically a summarization across the previous session. s615 00:45:31.119 --> 00:45:39.119 Just uh make it a lot easier uh for me to um keep uh getting a very clear sense of what is the next step. s616 00:45:39.119 --> 00:45:49.599 Yeah, I think for now I just want to wait for at least one PR that went through no mistakes pipeline so I can show you what does a PR that went through no mistakes s617 00:45:49.599 --> 00:45:53.040 uh and has done the more rigorous validation looks like. s618 00:45:53.040 --> 00:45:56.560 Uh it's very different from the YOLO PR that we merged earlier. s619 00:45:56.560 --> 00:45:59.200 Oh, here we can see something interesting is happening. s620 00:45:59.200 --> 00:46:00.800 the treehouse warning fix. s621 00:46:00.800 --> 00:46:09.359 Uh it's already validated uh pushed to this commit but we couldn't create a PR because GitHub seems to be returning internal server error. s622 00:46:09.359 --> 00:46:17.200 Um so first mate received this update and it basically authorized a retry asking the crew mates to just try again. s623 00:46:17.200 --> 00:46:19.040 Uh it seems it's healthy now. s624 00:46:19.040 --> 00:46:22.240 Um and gave me this updates and I didn't have to do anything. s625 00:46:22.240 --> 00:46:25.520 Um I didn't uh I may not even be looking at this message right now. s626 00:46:25.520 --> 00:46:33.040 So this is one of the values that you can get from firstmates which is that it's an orchestrator that will take care of these things for you. s627 00:46:33.040 --> 00:46:38.000 So you don't have to worry about oh there's a GitHub error and uh I need to retry. s628 00:46:38.000 --> 00:46:39.280 You don't need to think about that. s629 00:46:39.280 --> 00:46:41.359 First mate will just take care of it. s630 00:46:41.359 --> 00:46:44.160 Okay we have a new lavish board to review. s631 00:46:44.160 --> 00:46:46.000 Uh let's take a look. s632 00:46:46.000 --> 00:46:48.720 Um five kinds. s633 00:46:48.720 --> 00:46:52.319 This is a followup uh on the five birds. s634 00:46:52.319 --> 00:47:02.480 Um I think yeah so it's figuring out how do we let people choose the birds uh and it proposed uh all these different colors as well. s635 00:47:02.480 --> 00:47:06.079 Uh so each kind of bird can have many many different kind of colors. s636 00:47:06.079 --> 00:47:07.839 This is pretty cool. s637 00:47:07.839 --> 00:47:11.040 Um okay. s638 00:47:11.200 --> 00:47:11.599 Yeah. s639 00:47:11.599 --> 00:47:12.880 Looks nice, right? s640 00:47:12.880 --> 00:47:14.160 Looks really nice. s641 00:47:14.160 --> 00:47:15.040 Um okay. s642 00:47:15.040 --> 00:47:16.880 So this is pretty straightforward. s643 00:47:16.880 --> 00:47:18.480 I would say we just do it. s644 00:47:18.480 --> 00:47:20.319 Um the word opens. s645 00:47:20.319 --> 00:47:21.040 Uh oh. s646 00:47:21.040 --> 00:47:22.160 How how does it open? s647 00:47:22.160 --> 00:47:23.440 Okay. s648 00:47:23.440 --> 00:47:29.119 Um I think that's pretty straightforward. s649 00:47:29.760 --> 00:47:31.599 Yeah. s650 00:47:31.599 --> 00:47:34.880 Um let's see what decisions do we have. s651 00:47:34.880 --> 00:47:38.960 Oh, it's even made sure um the UI will work well on the phone. s652 00:47:38.960 --> 00:47:41.760 Um that's very thorough. s653 00:47:41.760 --> 00:47:43.599 Uh cool. s654 00:47:43.599 --> 00:47:47.839 So decisions, recommendations, uh decisions for the captain. s655 00:47:47.839 --> 00:47:50.720 What does the corner word do? s656 00:47:50.720 --> 00:47:52.240 Uh, opens the strip. s657 00:47:52.240 --> 00:47:53.359 Yeah. s658 00:47:53.359 --> 00:47:57.119 Uh, grouped by kind on colors first. s659 00:47:57.119 --> 00:47:59.680 Keep cycling kinds. s660 00:47:59.680 --> 00:48:04.960 Yeah, I think we should do a Yeah, just a straightforward chooser. s661 00:48:04.960 --> 00:48:08.800 Whose colors fly when nobody chooses? s662 00:48:08.800 --> 00:48:10.480 Each kind has its own color. s663 00:48:10.480 --> 00:48:12.000 Yeah, that's good. s664 00:48:12.000 --> 00:48:14.640 Uh, what should be promoted from this scout? s665 00:48:14.640 --> 00:48:21.200 Uh we want uh I want to build this um and together with the list under Yeah. s666 00:48:21.200 --> 00:48:22.079 everything. s667 00:48:22.079 --> 00:48:22.800 Okay. s668 00:48:22.800 --> 00:48:24.079 Uh send it back. s669 00:48:24.079 --> 00:48:25.760 Let's build it. s670 00:48:25.760 --> 00:48:28.960 Uh oh, another board, the Snowy Mountains. s671 00:48:28.960 --> 00:48:29.520 Okay. s672 00:48:29.520 --> 00:48:31.680 Uh that's also ready for review. s673 00:48:31.680 --> 00:48:32.079 Okay. s674 00:48:32.079 --> 00:48:36.240 So, it looked at some real photos of uh Mount Everest, right? s675 00:48:36.240 --> 00:48:37.680 Uh that's what I asked. s676 00:48:37.680 --> 00:48:40.240 And it's got these reference images. s677 00:48:40.240 --> 00:48:47.359 Uh it like this made sure the agent actually understood what I mean by like goodlooking mountains. s678 00:48:47.359 --> 00:48:50.559 Um so this is um good. s679 00:48:50.559 --> 00:48:54.400 Uh so uh okay. s680 00:48:54.400 --> 00:49:01.119 Uh oh yeah, this is starting to look better because it's got like the rocks and um the snow. s681 00:49:01.119 --> 00:49:04.400 Um Oh yeah. s682 00:49:04.559 --> 00:49:07.440 Um okay. s683 00:49:08.319 --> 00:49:11.359 What else do we need to review? s684 00:49:11.359 --> 00:49:12.160 Um, okay. s685 00:49:12.160 --> 00:49:14.640 So, these are the previous variants, right? s686 00:49:14.640 --> 00:49:22.640 And I think, uh, the agent just introduced this new variants D, uh, here. s687 00:49:22.640 --> 00:49:24.079 Uh, okay, cool. s688 00:49:24.079 --> 00:49:25.359 I think this is better. s689 00:49:25.359 --> 00:49:29.040 Um, so let's probably make a decision on this as well. s690 00:49:29.040 --> 00:49:33.359 Um, compare. s691 00:49:35.200 --> 00:49:36.319 Uh, okay. s692 00:49:36.319 --> 00:49:37.359 Let's see. s693 00:49:37.359 --> 00:49:40.240 Uh, I think the variant D is the new one. s694 00:49:40.240 --> 00:49:44.319 Um, and yeah, I think it's like better than the other ones. s695 00:49:44.319 --> 00:49:48.640 Um, so let's do it. s696 00:49:48.640 --> 00:49:49.920 Uh, let's do it. s697 00:49:49.920 --> 00:49:50.880 Let's do it. s698 00:49:50.880 --> 00:49:55.359 Um, D. Wait. s699 00:49:55.359 --> 00:49:58.720 Uh, yeah, it's just straightforward D, right? s700 00:49:58.720 --> 00:49:59.200 Uh, okay. s701 00:49:59.200 --> 00:50:04.559 Let's just choose D. Should the mountains themselves change? s702 00:50:04.559 --> 00:50:07.760 Uh what? s703 00:50:07.760 --> 00:50:09.599 How does the mountain change? s704 00:50:09.599 --> 00:50:11.920 Pyramid peaks as shown. s705 00:50:11.920 --> 00:50:13.200 Uh taller. s706 00:50:13.200 --> 00:50:15.760 Yeah, I do want the shapes to change. s707 00:50:15.760 --> 00:50:20.960 Uh taller and sharper. s708 00:50:20.960 --> 00:50:24.000 Um okay, cool. s709 00:50:24.000 --> 00:50:25.200 That's it as well. s710 00:50:25.200 --> 00:50:27.760 So yeah, this is basically how I make decisions. s711 00:50:27.760 --> 00:50:37.839 uh the agents will bring these decisions to me with a lot of context with a lot of the um uh the homework that I can see um and the visuals the prototypes as well. s712 00:50:37.839 --> 00:50:48.960 This process really helps me understand what the agent has done what we are actually choosing between and also helps the agent get to a good implementation that I will be happy with as well. s713 00:50:48.960 --> 00:50:50.800 Uh so let's get back to this. s714 00:50:50.800 --> 00:50:53.200 Um yeah the choices are locked in. s715 00:50:53.200 --> 00:51:03.440 I think I might have made a mistake in the beginning and I didn't actually get the northern lights uh prototype to start uh because there's no crew mates that's doing the northern lights. s716 00:51:03.440 --> 00:51:05.440 Uh yeah, I guess that's okay. s717 00:51:05.440 --> 00:51:06.319 I'll do that later. s718 00:51:06.319 --> 00:51:08.640 Uh for now uh we'll just ignore that. s719 00:51:08.640 --> 00:51:10.800 Okay, we just have two more PRs landed. s720 00:51:10.800 --> 00:51:14.800 Uh the bird varants uh selector and the treehouse warning fix. s721 00:51:14.800 --> 00:51:17.200 Uh let's look at the treehouse one. s722 00:51:17.200 --> 00:51:19.760 Um because I think this one went through no mistakes. s723 00:51:19.760 --> 00:51:26.559 Uh and I also let um first mate do yolo merge for this one because treehouse does not automatically release. s724 00:51:26.559 --> 00:51:29.680 Uh every time I create a new release I will test things out as well. s725 00:51:29.680 --> 00:51:35.599 Um but even at this PR level um no mistakes does a lot of heavy lifting for us. s726 00:51:35.599 --> 00:51:37.760 Uh so I'll just walk through the PR structure here. s727 00:51:37.760 --> 00:51:41.760 The intent was basically my requirements uh for this task. s728 00:51:41.760 --> 00:51:47.359 Um and what changed there's a quick summary uh uh the duplicated uh warning. s729 00:51:47.359 --> 00:51:47.839 Yep. s730 00:51:47.839 --> 00:51:49.359 Uh that's what we asked. s731 00:51:49.359 --> 00:51:51.280 Uh and there's a risk assessment. s732 00:51:51.280 --> 00:52:03.839 This is very helpful because um usually when the PR is having a uh big blast radius or having some kind of unresolved uh problem, it will have a medium or high uh risk assessment for me. s733 00:52:03.839 --> 00:52:08.800 When I see this is low, I already know like I don't I probably don't need to take too much look. s734 00:52:08.800 --> 00:52:10.720 Um and there's a testing section. s735 00:52:10.720 --> 00:52:17.040 Uh these are the live validation that um no mistakes the agents in the no mistakes pipeline did. s736 00:52:17.040 --> 00:52:20.319 Uh so it did uh nine out of 10 scenarios. s737 00:52:20.319 --> 00:52:27.440 Uh and these scenarios are basically the things that prove this change does work according to our intent. s738 00:52:27.440 --> 00:52:27.760 Right? s739 00:52:27.760 --> 00:52:37.920 So uh it said get this um uh this lease will wants on std error um repository level hooks remain untrusted uh not executed. s740 00:52:37.920 --> 00:52:38.160 Yeah. s741 00:52:38.160 --> 00:52:44.160 So it basically like validated everything that we care about this live with the product end to end. s742 00:52:44.160 --> 00:52:46.160 Uh and there's evidence as well. s743 00:52:46.160 --> 00:52:55.040 Um and I think there's like yeah there is further like uh logs that we can look at to uh understand uh what was really tested. s744 00:52:55.040 --> 00:53:04.240 There was one scenario skipped uh and this one is untested uh repository config configuration documentation directs user to places. s745 00:53:04.240 --> 00:53:08.640 Oh documentation okay yeah documentation does not need a live testing. s746 00:53:08.640 --> 00:53:10.000 So uh that's probably it. s747 00:53:10.000 --> 00:53:10.240 Yeah. s748 00:53:10.240 --> 00:53:20.319 So this risk assessment and testing uh this validation is basically how I develop uh confidence in a code change without having to look through every single line of the code. s749 00:53:20.319 --> 00:53:25.200 Usually when the risk assessment is not low then I will go take a look at the code as well. s750 00:53:25.200 --> 00:53:28.640 Uh but when it's low and it's well tested I don't even look at the code. s751 00:53:28.640 --> 00:53:30.079 Uh it's usually just good. s752 00:53:30.079 --> 00:53:30.640 Okay. s753 00:53:30.640 --> 00:53:34.640 We are probably not going to wait for every single uh crew mate to finish. s754 00:53:34.640 --> 00:53:43.760 Hopefully this video so far has already given you a look into how I work, how I direct a large number of agents to work in parallel without going insane. s755 00:53:43.760 --> 00:53:47.520 I will drop a link to everything I mentioned in the description of this video. s756 00:53:47.520 --> 00:53:53.440 And if you have any questions, please just leave a comment or join my Discord server and I'll try to get back to you. s757 00:53:53.440 --> 00:53:57.200 Thank you for watching and see you next time.