High Throughput Agentic Engineering with ex-Meta L8 Principal

Kun Chen · 2026-09-13 · 53 min · 757 sentences · from YouTube's caption track

Each timecode opens YouTube at the start of that sentence. Line anchors (#s42) are the cue ids in the WebVTT, and every line carries its start and end seconds. All transcripts has every talk, and the whole corpus as one file.

1. 00:00Hi everyone, or should I say hi captains.
2. 00:03Many of you might have already watched my previous videos because they got more than a million views.
3. 00:08If you haven't yet, my name is Kun.
4. 00:10I was an L8 principal engineer previously at Meta, Microsoft, and Atlassian.
5. 00:15My previous videos shared a lot of the basics about agentic engineering.
6. 00:20So this time I'm hoping to share a more advanced workflow focusing on high throughput multitasking through a large number of agents.
7. 00:29I'll walk through it by just showing you how I get some real work done.
8. 00:33Let's jump in.
9. 00:34As always, we will start with a clean terminal.
10. 00:37And the first thing I'm going to launch is Herder.
11. 00:39Herder is basically my new T-Max.
12. 00:41Uh it's a multiplexer.
13. 00:42It manages all my agent sessions and allows me to easily navigate between the sessions.
14. 00:47You will see that in a bit in this video.
15. 00:49And then I'm going to go into my first mate directory.
16. 00:53Uh 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.
17. 01:02This is basically my workflow.
18. 01:04First 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.
19. 01:20uh is first mate doing that and you will see that in a bit.
20. 01:23Um the way to set up first mate is super easy.
21. 01:25You just just clone this repo and launch your agent in this repo.
22. 01:29Uh so now that I'm in the first mate directory, I just launch my agent.
23. 01:33Uh I use pi.
24. 01:34Um I I use both pi and cloud code.
25. 01:37Uh I basically use cloud code for whenever I want to use an anthropic model uh like opus and fable.
26. 01:43And I use pi for everything else because pi uh basically it can use any other model.
27. 01:48Uh allows me to switch between models very easily as well.
28. 01:52Uh 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
29. 02:05is uh faster and more efficient and more straight to the point.
30. 02:08Uh 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.
31. 02:18Um so I like 4.5 better.
32. 02:20Uh 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.
33. 02:28Uh 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.
34. 02:36Uh so yeah, this is first mate and this agent session is where I literally do everything.
35. 02:43All my projects, all my tasks, I just use this single agent session to do it.
36. 02:47And maybe to show you what that looks like, I'll just ask first mate list all the projects that we manage through this fleet.
37. 02:57Um yeah, and you notice that I use voice input by default.
38. 03:01Uh and that's how I work.
39. 03:02uh voice input is just faster than uh typing uh most of the time like this kind of natural language prompts.
40. 03:08Uh the only case where I don't use voice is when I have to um type a URL or something like that.
41. 03:14Yeah.
42. 03:14And you can see first mate has already listed all the projects that I currently manage through first mate.
43. 03:19Uh it's a lot of projects.
44. 03:20Uh right uh this is like how many 36 projects in total.
45. 03:24Um so this is uh I do everything uh across these 36 projects.
46. 03:28I do a lot of tasks.
47. 03:29Uh most of these are open source projects that I uh published.
48. 03:33Some of them are private projects as well.
49. 03:35And I do everything just through first mates.
50. 03:37And to show you how I do that, uh maybe I just like walk through some real tasks.
51. 03:41Uh the first thing I probably want to do is u this project called backpass.
52. 03:46Um backpass is a um tool that you can use to help you improve your agent um agents.mmd or cloth.md.
53. 03:55The 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
54. 04:06so that your agents can work better next time.
55. 04:09Um, and this tool uh I'll probably do a dedicated video about this in the future.
56. 04:13Uh, but right now I just want to do one improvement which is uh something I'll just tell first mate.
57. 04:20Hey first mate, previously we planned an idea for backpass which is to allow backpass to collect agent sessions across multiple remote machines.
58. 04:29Um uh I think the plan is already there.
59. 04:32Can we pull that out and then dispatch a crew mate to work on that?
60. 04:38All right.
61. 04:39Uh so that's how I uh ask first mate to do one task.
62. 04:43Uh 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
63. 04:56uh agent to do it.
64. 04:58Um you will notice that um first made agent is already working.
65. 05:02Um but we don't see any tool calls.
66. 05:04We don't see anything in in this uh pi agent terminal, right?
67. 05:07Um 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.
68. 05:18Um this is done by a custom PI extension that I built.
69. 05:23Uh uh it's uh it comes with a first mate repo.
70. 05:26Um and the reason I do this is that I really don't need to look at what's the agent is doing.
71. 05:32It's making all those tool calls.
72. 05:34It's reading files.
73. 05:35It's doing this and that.
74. 05:36um I don't actually need to know that most of the time those things are distractions and noise.
75. 05:42Uh 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
76. 05:54and just show this little boat that I uh can allow me to know the agent is working.
77. 05:59It's not stuck.
78. 06:00Um, 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.
79. 06:09So now I can see, oh, it's making this tool call.
80. 06:11It's uh spawning a crew mate, right?
81. 06:14Um so yeah, that's what the agent is doing.
82. 06:16But most of the time I just turn on com mode and I just uh uh don't need to look at all those noise.
83. 06:23Um and then that allows me to focus.
84. 06:25that allows me to actually think about what do I actually need to do next, right?
85. 06:30Uh so that's what's important in not looking at the tool cost and everything.
86. 06:34Um 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.
87. 06:44Um 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.
88. 06:52Uh 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
89. 07:02uh 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
90. 07:12um 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
91. 07:25uh tool calls.
92. 07:27Um we can probably see that um before first mate decided to um launch the crewmate, it caught a few things.
93. 07:36It called this quotota axi uh and it used a skill called quotota array dispatch.
94. 07:42Uh this is something that's really important.
95. 07:44Uh and uh that's why I want to talk through this.
96. 07:47Um, 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.
97. 08:00Um, so let me uh turn this back off and uh let me show you and let me go find the config file.
98. 08:06Uh, it's a config file in the config directory called crew dispatch.json.
99. 08:11Um, this is my dis uh dispatch rules.
100. 08:14You can configure yours.
101. 08:15Um, and I actually don't write this rule file by by hand.
102. 08:19I ask first mate to uh do this.
103. 08:22I just tell firstmate what my preferences are and first mate will configure this file for me.
104. 08:27Um, 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.
105. 08:35Um, Fable is like really good at building good UI.
106. 08:39Um, and this iOS app actually uh is a paid app.
107. 08:42I really want to make sure it's taken good care of.
108. 08:44Uh so I use Fable for that.
109. 08:46Um and the task requires uh generating images.
110. 08:51Uh if it needs images then I use codeex as the harness and then GPT 5.6 soul.
111. 08:56Um codeex has a native tool for generating images because openai has the uh image generation model right.
112. 09:04Uh so I use codeex to do that.
113. 09:06Um there is also uh when the task is technical product design architecture um or planning work that is like genuinely very difficult.
114. 09:15Uh so it's a difficult planning work.
115. 09:17I use fable I use uh Kim K3 I use Astra.
116. 09:22Uh so I basically allow these three models to do this kind of very complex and ambiguous planning for me.
117. 09:28Uh and there's like Y as well.
118. 09:30Uh so it helps the model uh make the decisions.
119. 09:33uh 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
120. 09:43cursor 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.
121. 09:57Uh 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
122. 10:08there 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.
123. 10:22So I'll show you um how I look at my quota as well.
124. 10:26I use this tool called quota axi.
125. 10:28That's another project I built.
126. 10:29Uh it can give us um this data uh where I can just easily monitor how my quota is doing.
127. 10:36I have a subscription from cloud.
128. 10:38I have a codeex.
129. 10:39I have a cursor.
130. 10:40I have a gro as well.
131. 10:42Uh and this is where they uh sit currently.
132. 10:45Uh I actually took a vacation right before this and I have a ton of kota available uh which is awesome.
133. 10:51Uh 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.
134. 11:00Uh 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.
135. 11:12Uh so this is like really useful because otherwise I have to constantly look at this myself.
136. 11:17Um 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
137. 11:28uh 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
138. 11:41um 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.
139. 11:49Uh there is another project um that I want to uh make some improvements on.
140. 11:54Uh I'll show you.
141. 11:56There is a project I recently built.
142. 11:57I really uh find quite interesting.
143. 12:00Uh it's called Fry with Me.
144. 12:02Uh and this project uh is a uh it's a game basically.
145. 12:06Um and let me show you.
146. 12:08Uh I'll launch a new seed.
147. 12:09Uh it's basically a game, but it's kind of uh it's something to help me relax.
148. 12:14I 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.
149. 12:22Uh it's just like a really relaxing uh bird that is flying across the world.
150. 12:28Uh and this whole world is u procedurally generated.
151. 12:31Uh so it's always different.
152. 12:33Uh and I can just put this on a second monitor and like uh keep focusing on my work.
153. 12:38Um, 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.
154. 12:46Uh so I'll just go back to um my first mate and say uh tell first mate some of the ideas.
155. 12:54Hey um I want to make some uh improvements to uh fry with me.
156. 12:59Uh I have three ideas.
157. 13:01One is that I want to prototype uh some better looking snowy mountains.
158. 13:07Uh so when there's a mountain with snow uh right now it doesn't look as good.
159. 13:12Uh I want to prototype a few really goodlooking uh mountains um that we can uh render.
160. 13:19Uh so that's one.
161. 13:20Another idea I have is that uh at night right now uh we just show a uh night sky.
162. 13:27We have the Milky Way which is good.
163. 13:29Uh but I also want to have northern lights.
164. 13:32Um they should show up kind of rand randomly.
165. 13:35uh not always uh but uh they should look like uh realistic northern lights.
166. 13:40Um the third idea is that right now there's only one bird and we cannot choose between different kind of birds.
167. 13:46Uh I want to prototype uh a few different kind of birds that people can choose.
168. 13:51Um so yeah for each of these ideas I want you to um prototype three different variants.
169. 13:58Uh so I can choose uh which one uh I like.
170. 14:02Um and uh once the prototypes are ready, uh ask the crew mates to give me uh that in lavish for me to review.
171. 14:12And each idea should be done by a fable crew mate uh in parallel.
172. 14:20All right.
173. 14:20Uh so basically I gave a long ramble of uh ideas that I have.
174. 14:24Uh I like to do this a lot.
175. 14:26Um and you can see I asked for prototypes.
176. 14:29Um and I asked the prototypes to be given to me uh through lavish which we'll look at uh in a bit.
177. 14:36Uh it's a really good way to visually validate uh the ideas and just pick uh very easily and give feedback as well.
178. 14:42Um and I specifically asked for fable uh because uh the reason uh was like uh I was looking at the quota, right?
179. 14:50We have a whole bunch of fable quota available.
180. 14:53Uh so why not uh use that?
181. 14:54Uh so I use fable uh to do this prototypes uh and uh they usually uh can give really good results.
182. 15:00Yeah.
183. 15:01So um this is basically uh while the previous backpass uh task is still ongoing uh the crew mate is still working on that.
184. 15:09uh 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
185. 15:24and 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
186. 15:39more prompts and kill those prompts.
187. 15:41All right, so I don't actually have to wait for first mate to uh finish up or anything.
188. 15:46Uh I just keep giving more and more prompts.
189. 15:49Um and that's how I uh work with first mate uh to like get a lot of tasks spun up in parallel.
190. 15:56Um 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.
191. 16:07Uh right.
192. 16:08Um 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.
193. 16:16Um in first mate there is a built-in skill called bearings that can do this.
194. 16:21Uh this will basically allow first mate to collect a snapshot of all the uh active tasks across the fleet and give me a snapshot.
195. 16:30Um but uh the default bearings will give me that uh in the terminal.
196. 16:35Uh which sometimes is like a little bit hard to read.
197. 16:37Uh so I actually have a variant of that called bearings lavish which will get first mate to give me that report in lavish.
198. 16:45Uh so I'll just show you.
199. 16:46Um I used the follow-up prompt to cue this uh right u behind the current actions.
200. 16:51Uh so first mate will not run this yet.
201. 16:54Uh it's queued as a follow-up.
202. 16:56So only when first mate has done the current set of actions then it will uh do the follow-up prompt.
203. 17:03Uh so this is a good way to make sure first mate is not being distracted uh it's actually doing things one by one.
204. 17:08Um if I need to uh actually steer first mate uh to do something I can also steer instead of follow-up.
205. 17:14Uh so followup I think is alt enter and steer is just enter.
206. 17:19So if I type a prompt and enter it will become a steer that will immediately send to first mate.
207. 17:25Um that's a PI feature actually.
208. 17:27Um and uh it's very handy to have a distinction between follow-up and steer.
209. 17:32Um 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
210. 17:48uh 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
211. 18:01uh for the task task that I uh am about to do.
212. 18:04Um this is useful when I have some opinion or judgment on what is the right model.
213. 18:10Um and um yeah uh for different tasks I can just easily uh get the right model to work on them.
214. 18:16Uh and right right now uh first mate said all those three um fly with me prototypes uh underway.
215. 18:22Uh and now the follow-up prompt bearing stavish uh just uh got sent right.
216. 18:27Uh so that's how a follow-up prompt work.
217. 18:29All right.
218. 18:30First 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
219. 18:42that has uh what's uh what's charted next uh what is currently underway.
220. 18:46Uh 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.
221. 18:55Uh so let me look through this.
222. 18:57These were usually like things that were previously discussed uh and haven't been done.
223. 19:02So uh it's good to u make a decision on.
224. 19:05So there's a workaround for lavish onboard.
225. 19:08Uh this one yeah I remember this one but I don't want to do that right now.
226. 19:12Uh I just decide claude exit confirm dialogue.
227. 19:15Uh this is for first mate.
228. 19:17Uh 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.
229. 19:28Uh so I think we actually have to do this uh treat as non uh non wedge.
230. 19:34Yep.
231. 19:36Uh this le props up sub uh yeah this one I remember I I don't want to do that right now.
232. 19:41Um uh and I'll just look through like all these decisions.
233. 19:45These were usually like things that we previously discovered.
234. 19:48Uh herder missing end points recreate.
235. 19:50Yep, that's a bug.
236. 19:51Uh keep refuse escalate.
237. 19:55Um yep, that's the right behavior.
238. 19:59Uh URL wrap.
239. 20:00Uh this is for the ship app.
240. 20:02Um and this is a bug uh will uh we'll fix now.
241. 20:06Uh ship a follow-up.
242. 20:07Um this is a another improvement for ship uh for a uh first-time user experience uh to teach user about the multiplexer.
243. 20:16Uh this is quite complex.
244. 20:17Uh so I'm going to keep that held uh and discuss that a little bit deeper.
245. 20:21Uh this is the diaglide banner.
246. 20:24Um yeah, this one I actually want to build now.
247. 20:26Uh so I'll do that.
248. 20:28Uh design kit mock removal.
249. 20:30Yep, I remember this one needs to be done.
250. 20:32Uh I'll decide as well.
251. 20:35Um yeah, you can see there are quite some decisions, right?
252. 20:38Uh 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.
253. 20:49Um so uh it's good to have this like first mate keeping track of that for me.
254. 20:54Um dial nested fan editor.
255. 20:56Yep, I need that now.
256. 20:58Um what else?
257. 21:00Drag into group.
258. 21:02Uh yeah, this is uh where like um users can customize uh the dial in the ship app.
259. 21:09Um I will do that later.
260. 21:12Um so okay, so all the decisions uh we have recorded uh the follow-up.
261. 21:16Uh so I'll just send this back.
262. 21:18Yeah, we can see here this working spinner.
263. 21:20That means uh the feedback uh was sent back to first mate.
264. 21:24Uh and uh I can actually just end the session now.
265. 21:27Uh if I go back to first mate, yep, first mate is now um having a boat.
266. 21:31uh which means first mate is working.
267. 21:33Um so it's likely that first mate just received the feedback and is now acting on them.
268. 21:38Uh we'll give first mate a bit of time to see uh what first mate does.
269. 21:42Okay, here we can see uh first mate just got back to us.
270. 21:45Uh it said um the uh the four uh decisions about first mate uh is listed here.
271. 21:51Uh actually there's like nothing needed.
272. 21:53Uh the one thing I decided to build was actually already released.
273. 21:57It's already in the codebase.
274. 21:59Um so uh ship uh we decided to um build this now build this now and fix that and prioritize this.
275. 22:07Uh 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.
276. 22:15We can see there's a crew mate under here that's uh currently running.
277. 22:19Um second mate is basically a way to scale up uh first mate's charter.
278. 22:24Um 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
279. 22:32um 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
280. 22:48mates to own different domains um of the charter so these second mates uh ship uh owns everything related to this ship app.
281. 22:57So every uh piece of work that's on this ship app will get routed to this second mate.
282. 23:02So first mate doesn't have to orchestrate all that.
283. 23:05Uh it can just send the ask to this second mate and the second mate will do it.
284. 23:09Um 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.
285. 23:17Uh there's a few more that will likely be spun up soon as well.
286. 23:21Um, every second mate is actually a first mate by itself.
287. 23:25Um, so I can actually just go into this second mate and see what it's doing.
288. 23:29Uh, I can talk to it just like a normal first mate.
289. 23:32Um, and uh, maybe uh, let me show one example.
290. 23:35Uh, this is uh, a second mate for the Edis wallets app.
291. 23:39Uh, this app was the app I built in my last video.
292. 23:42This app is actually already released in the app store.
293. 23:45U, but I actually forgot uh, which features got in versus not.
294. 23:49So, I'll just ask uh which features have already landed that's not released to the app store yet, right?
295. 24:00Um so I can just talk to this second mate.
296. 24:02Every second mate is also a first mate.
297. 24:04Uh so it will manage its own crew mates.
298. 24:06Uh it has its own memory and uh instructions and everything.
299. 24:10Um so I can just talk to it.
300. 24:12Uh I can also talk to it through first mate, right?
301. 24:16I can ask first mates the same question here and first mate will route the question to that second mate uh whenever appropriate.
302. 24:23Most of the time uh I actually go with the easy route where I just talk to first mate.
303. 24:28Yeah.
304. 24:28So this second mate is now uh trying to answer our question uh and we can let it do its work.
305. 24:33Um and you can see here um there is a local versus mini.
306. 24:38Uh this is a herder feature.
307. 24:40So Herder has the ability to manage all the agent sessions across multiple machines.
308. 24:46Um so this mini is my Mac Mini uh which is uh uh actually on my shelf right now.
309. 24:52Um it's running headlessly.
310. 24:54Uh so I don't connect any monitors or mouse or keyboard to the Mac Mini.
311. 24:58I just let it run uh headlessly.
312. 25:00And this mini actually manages a lot of uh have a lot of second mates running on it.
313. 25:05Uh 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.
314. 25:14Um so the mini uh has uh basically more hardware resources than my MacBook.
315. 25:20This current local machine is a MacBook.
316. 25:22Uh it has uh a little bit less power uh than what my mini has.
317. 25:26Uh so I assigned a lot of domains uh to the Mac Mini uh so it can do more work.
318. 25:32Okay.
319. 25:32Uh there's actually a lavish board that's just brought up to me.
320. 25:35Um and we can see it's the bird kinds for fly with me.
321. 25:39Uh this is what I asked earlier, right?
322. 25:41Uh and it's done by Fable.
323. 25:43Uh so now it's ready for review.
324. 25:45Um it's really nice to have this lavish uh interactive artifact.
325. 25:48You will see um the current bird, right?
326. 25:51This is the current bird.
327. 25:52It took screenshots so that I can see what's the current bird looks like.
328. 25:56Uh and board soaring raptor.
329. 26:00Okay, this is the prototype, right?
330. 26:01So, this is a new kind of bird.
331. 26:03Uh, and it's giving me this preview.
332. 26:05Uh, this is really nice.
333. 26:07Uh, I I really like um looking at this previews.
334. 26:11Um, so I actually like this one.
335. 26:13Uh, this this is okay.
336. 26:14Uh, what is this slender swallow?
337. 26:18Uh, what is a what is a slender swallow?
338. 26:20I I don't know either of those words.
339. 26:23Um, but uh it it looks nice.
340. 26:26Uh, it's okay.
341. 26:27This one I probably will also keep.
342. 26:29Uh there is a large ocean glider.
343. 26:32Okay, this is just like the current one, right?
344. 26:35Like it's uh a little bit different but very similar.
345. 26:38Okay, there's a long necked crane.
346. 26:42Oh, cranes.
347. 26:43I like cranes.
348. 26:44Uh they are good.
349. 26:46What is this soft night owl?
350. 26:48Wow, it has so many prototypes.
351. 26:50It's done a lot of work for us.
352. 26:52Um so, okay.
353. 26:53Uh so there's an owl.
354. 26:54I kind of like this owl as well.
355. 26:56It's like round and cute.
356. 26:58Um, so, uh, what do we need to look at here?
357. 27:02Each kind closer.
358. 27:03Okay, closer.
359. 27:04Uh, okay, that's closer.
360. 27:07Um, it actually explained, uh, like the key features, um, of each, um, bird.
361. 27:12Uh, that's really nice.
362. 27:14Uh, but I think I already like have seen enough about these birds to know what I like.
363. 27:19Uh, usually at the bottom it will list uh, the decisions for me to make, right?
364. 27:23So, recommendations.
365. 27:24Uh, carry forward, carry forward.
366. 27:26Uh 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
367. 27:43broad 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.
368. 27:57Uh I need the crane.
369. 27:58I need the owl.
370. 27:59Okay.
371. 28:00Uh so yeah, these are our decisions.
372. 28:03And we'll just kill this uh and send it back.
373. 28:06Uh I'll just say like uh let's build uh the birds we decided.
374. 28:12Okay.
375. 28:13And send it an end.
376. 28:14And that's it.
377. 28:15That's my feedback back to Fable.
378. 28:17and Fable um that crewmate who uh prototyped this should uh receive my feedback about which birds that we wanted to forward.
379. 28:26Um and uh that's it.
380. 28:28Uh so that's the um only thing we had to do with that prototype.
381. 28:33Um we can see here uh ship has already spun up three ship crew mates, right?
382. 28:38Uh working on three things in parallel.
383. 28:40So right now how many do we have?
384. 28:41We have one, two, three, four, uh five, six, seven.
385. 28:45We have seven crew mates working in parallel.
386. 28:47Um, and we can keep giving more.
387. 28:49Like first mate is idle right now, right?
388. 28:51First mate is not doing anything.
389. 28:52We can just keep giving more ideas.
390. 28:54Um, I uh that's what I will typically do.
391. 28:57I just keep giving more and more.
392. 28:59Um, and this edi wallet just answered our question as well.
393. 29:03The second mate, uh, it landed since this not released.
394. 29:07Uh, there's a few Okay, there's a few new userfacing features that's landed but not released.
395. 29:12Uh, auto pay local.
396. 29:14Oh, these are not released yet.
397. 29:15Okay.
398. 29:16Uh yeah.
399. 29:17So now I can decide uh do I want to like release those features.
400. 29:20Uh but I'll probably put that aside for now.
401. 29:23Uh we have enough uh to like work through here.
402. 29:26Yeah.
403. 29:26So 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.
404. 29:39Uh 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.
405. 29:46Um so um we have um like constant discussions about first made about uh models and harnesses uh as well.
406. 29:53Uh 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.
407. 30:03Uh 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.
408. 30:12Um let me see uh what I haven't looked at yet.
409. 30:15Um I think this one I um probably haven't looked at yet.
410. 30:19So this one uh what is this?
411. 30:21Um uh it's a treehouse.
412. 30:23Okay.
413. 30:23Treehouse is another project I have.
414. 30:25Uh and uh it's a bug report saying this uh hook is not firing.
415. 30:30Okay.
416. 30:31So uh it seems there's a lot of debug information already available.
417. 30:35So something I will show you here is that I can actually add my first mate.
418. 30:39Uh so I can just uh there's a bot here.
419. 30:41Uh that's my first mate.
420. 30:43I can just ask my first mate uh to uh investigate this, right?
421. 30:48Uh so I can just uh add my first mate here.
422. 30:51And this bot is actually relay uh relaying this uh message back to my actual first mate here running this local session.
423. 31:01Uh so this uh boat that's moving is likely uh from that mention.
424. 31:06So u I can just like basically summon my first mate uh this single agent session that I have from different places.
425. 31:14Uh I can right now it supports discord and uh x as well.
426. 31:18Uh maybe I can also show you that.
427. 31:20Uh 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
428. 31:33uh 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.
429. 31:52All right.
430. 31:53Um, yeah.
431. 31:54So, this at uh this mention will also get relayed to my first mate session.
432. 32:00Yeah.
433. 32:00So, now if we come back to this Discord thread, uh we should see first mate reply uh very soon.
434. 32:06Now, yeah, it just replied uh I captain looking into the treehouse predestroy hook uh not firing on destroy.
435. 32:12Uh we'll report back.
436. 32:14Yeah.
437. 32:14So now if we come back to this X thread, uh first mate also replied uh saying I captain clarifying that send feedback version message.
438. 32:20Uh right.
439. 32:21So uh that's really good.
440. 32:22Uh this is just like how I uh work across different surfaces.
441. 32:26It's so nice to have all these messages from all these different channels and surfaces all funneled into this single agent session.
442. 32:34Right?
443. 32:34So this first made agent session literally holds all my intent and all the context across every project that we have.
444. 32:41Uh so it can play a really good role in coordinating across everything and orchestrating all the crew uh to get things done for us.
445. 32:49Okay, 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.
446. 32:57Um so here you can see um the Gro 4.5 uh the default context window limits is 500k tokens.
447. 33:04So here it says we already used more than 40% of that.
448. 33:07Um 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.
449. 33:22Um 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.
450. 33:31Um, sometimes you can get the agent to use less context to do the same things uh and get things done with less cost.
451. 33:38Uh, 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.
452. 33:50Um, so there's a lot of things you can do.
453. 33:51Um, but my take is that I actually advise against uh managing that yourself manually.
454. 33:58The reason is that our most precious resource right now, the biggest bottleneck across everything we do with agents is our attention and time.
455. 34:07Uh our human attention and time.
456. 34:09So 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
457. 34:21and that is actually the most important thing.
458. 34:24Um so I actually think you can just waste a little bit.
459. 34:27It's okay to not be optimal about everything, right?
460. 34:31Um there's um there's going to be tools, there's going to be things that will help us manage that automatically.
461. 34:37Um 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.
462. 34:45So right now uh I actually suggest what you do is that you set a constant threshold.
463. 34:51Um so 500k for grock is actually a very reasonable threshold.
464. 34:55Um 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.
465. 35:04That is a little bit too high I would say.
466. 35:07Um so uh cloud code uh there's different settings.
467. 35:10I think there's a environment variable you can set to uh limit your context window size.
468. 35:15Uh let me see if I can find it actually.
469. 35:17Um I have that in my dot files.
470. 35:20uh 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
471. 35:34um 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
472. 35:48uh the session so it doesn't become too expensive and too uh inefficient.
473. 35:53Um otherwise just forget about it.
474. 35:55Don't think about the context window too much.
475. 35:57If your brain is thinking about the context window all the time, you don't have enough time to think about what to build.
476. 36:03Um 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.
477. 36:12If you do that right, you will get back much much more value than what you can save with this little tricks.
478. 36:19Uh I hope that makes sense.
479. 36:21Oh, and here uh actually first mate reported uh the treehouse finding right.
480. 36:26So uh treehouse investigation is done.
481. 36:28Finding is here not a destroy bug.
482. 36:30Uh real defect is hooks uh intentionally discarded for safety.
483. 36:34What?
484. 36:35Uh and that drop is silent.
485. 36:38We dropped hooks.
486. 36:39Uh okay.
487. 36:42Oh, so hooks in the repo is being discarded.
488. 36:46Uh destroy never run.
489. 36:48Keep ignoring repo hooks.
490. 36:50Uh I now remembered that is a security thing.
491. 36:53Yes.
492. 36:54Um so emit a one-time warning.
493. 36:57Um okay.
494. 36:59Uh yes.
495. 37:01Uh let's ship the warning fix and also reply uh to that Discord thread uh for what is happening here.
496. 37:10Um yeah so uh we'll do that uh and that is the closure for the treehouse uh buck report.
497. 37:17Um so there there's a lot of bug reports coming in discord uh in GitHub issues.
498. 37:22Uh 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.
499. 37:30Uh 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.
500. 37:37I just talk to first mate and first mate will orchestrate everything for me.
501. 37:41Uh I don't have to like actually juggle between all these sessions.
502. 37:44uh 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.
503. 37:52Oh, first mate gave another update.
504. 37:53Uh so um there is this uh treehouse warning fix is dispatched.
505. 37:58Uh discord update is posted.
506. 38:00Um fry with me bird clients merged.
507. 38:02Okay.
508. 38:03Uh if you missed it, I did miss it.
509. 38:05Uh so there's a PR and we can see uh this PR didn't even ask for my review.
510. 38:11Uh and I'll show you why.
511. 38:12um is because uh I told first mate to do do that.
512. 38:15Uh I'll show you the rules.
513. 38:17Uh so in first mate uh I think it's in config uh config or in data in data um projects uh MD.
514. 38:31Uh so in this project MD there is a policy being set on every single project.
515. 38:36Uh so if I search for fly with me uh this is uh policy that I set is called direct PR plus yolo.
516. 38:44Um 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.
517. 38:57Um 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
518. 39:08u no one is actually depending on it.
519. 39:10So 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.
520. 39:21The ones that have YOLO on is usually the projects where a code change will not get released immediately.
521. 39:28Um and we have a chance to test them out before we uh release them.
522. 39:32Uh 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.
523. 39:44Um so this is very um helpful for managing different levels of uh guardrails.
524. 39:50Uh not everything need no mistakes.
525. 39:52Not everything need uh a uh human review.
526. 39:54Uh depending on how high stake that is.
527. 39:57Uh 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.
528. 40:12Uh that takes a long time and use a lot of tokens.
529. 40:15So I don't actually always use no mistakes uh for every change I do.
530. 40:19A good way to think about this is like for this code change, would you have asked a another human to review this code?
531. 40:26Uh if you would not have have another code reviewer, then you probably don't need no mistakes.
532. 40:31Um 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.
533. 40:43Um, 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.
534. 40:50Um, there's actually another uh ship uh change that's landed as well.
535. 40:54Uh, the design kit mock removal.
536. 40:56Uh, good.
537. 40:57That's good.
538. 40:58Um, I think we just had another uh lavish artifact ready for review as well.
539. 41:04This is the snow capped mountains uh that I asked for.
540. 41:06Uh, I'm very excited to look at what this looks like.
541. 41:10Uh, what the snow does today.
542. 41:12Okay.
543. 41:12Yeah, I don't need to know that.
544. 41:14Uh 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?
545. 41:22That's the reason I asked for a better snowy mountain.
546. 41:26Uh Fable Fable is like really uh careful here.
547. 41:30It's like done some calculations and everything.
548. 41:33Uh and it's it's got a good like climate model.
549. 41:37Uh interesting.
550. 41:39um three shading studies.
551. 41:41I did a lot of studies.
552. 41:42That's why it took so much time.
553. 41:44So there is this like painted alpine.
554. 41:47Okay.
555. 41:48There's this uh story book cap.
556. 41:50Okay.
557. 41:51Ah interesting.
558. 41:52Uh uh there's this glacial jade.
559. 41:57Uh how's this different from the other ones?
560. 42:00It looks like kind of similar.
561. 42:02Uh and uh what's this?
562. 42:05Carved peaks.
563. 42:06The terrain study.
564. 42:08Oh.
565. 42:09Okay.
566. 42:11Uh oh, this looks good.
567. 42:14Uh this is uh like it'll go above the cloud.
568. 42:16Uh that is also nice.
569. 42:20Uh compare.
570. 42:21We can actually um preview these different uh variants.
571. 42:29This is so nice.
572. 42:30like um uh it's just really good to have this visual uh validation for what this is actually going to look like.
573. 42:39Uh it's a really good way to discuss prototypes.
574. 42:42Um I actually quite like uh these curved peaks.
575. 42:45Uh this uh look pretty nice.
576. 42:47Um but I don't uh want to accept this just yet.
577. 42:52Um I think I actually want this uh the mountain peaks to look more like Mount Everest.
578. 42:59that kind of mountain if you know what I mean.
579. 43:01Um so I will uh give some feedback here.
580. 43:06Uh this is mostly good.
581. 43:08This is a good improvements.
582. 43:09Uh but I actually want the mountain peaks to look more like Mount Everest.
583. 43:14Uh 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
584. 43:28All right.
585. 43:29Uh, so that feedback goes back to the fable crew mates.
586. 43:31Uh, so that's my feedback.
587. 43:33Um, and let's, uh, come back here and we can see, uh, treehouse warning fix, uh, is also underway.
588. 43:40Maybe one thing I'll show you is like, uh, you can see like we went to the lavish artifact to review stuff, right?
589. 43:46Uh, and while we are reviewing, uh, the first mate has already like printed a lot of information.
590. 43:52Uh, right?
591. 43:52It's got so many things.
592. 43:54uh and across this whole session, did I miss something, right?
593. 43:58Did I actually look through everything First Mate told me?
594. 44:01It's really unclear.
595. 44:02And this is where the chat UX is a little bit limiting.
596. 44:05Um so what I have here is like in First Mate repo, there is a built-in skill called AOY.
597. 44:12It's just a really good way to catch up on things.
598. 44:15Um so a Hoy basically tells First Mate to summarize what First Mate has told you since your last message.
599. 44:22and uh also pull out any open decisions that you haven't made yet.
600. 44:26Right?
601. 44:27So 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
602. 44:36uh and there's still some things waiting for me.
603. 44:38Uh and these things uh I actually directly gave feedback uh to the crew mate.
604. 44:43So uh it should be okay.
605. 44:45Uh there's an open call uh snowy mountains.
606. 44:47Yep.
607. 44:48Uh 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
608. 45:00and 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.
609. 45:13Uh so this is basically a good way for uh me to get a sense uh did I miss something?
610. 45:19Is everything okay?
611. 45:20Uh just a hoy and I can just keep spamming a hoy.
612. 45:23It's a very cheap uh call.
613. 45:25It's this skill doesn't cost too many tokens.
614. 45:27Uh it's basically a summarization across the previous session.
615. 45:31Just uh make it a lot easier uh for me to um keep uh getting a very clear sense of what is the next step.
616. 45:39Yeah, 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
617. 45:49uh and has done the more rigorous validation looks like.
618. 45:53Uh it's very different from the YOLO PR that we merged earlier.
619. 45:56Oh, here we can see something interesting is happening.
620. 45:59the treehouse warning fix.
621. 46:00Uh 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.
622. 46:09Um so first mate received this update and it basically authorized a retry asking the crew mates to just try again.
623. 46:17Uh it seems it's healthy now.
624. 46:19Um and gave me this updates and I didn't have to do anything.
625. 46:22Um I didn't uh I may not even be looking at this message right now.
626. 46:25So 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.
627. 46:33So you don't have to worry about oh there's a GitHub error and uh I need to retry.
628. 46:38You don't need to think about that.
629. 46:39First mate will just take care of it.
630. 46:41Okay we have a new lavish board to review.
631. 46:44Uh let's take a look.
632. 46:46Um five kinds.
633. 46:48This is a followup uh on the five birds.
634. 46:52Um 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.
635. 47:02Uh so each kind of bird can have many many different kind of colors.
636. 47:06This is pretty cool.
637. 47:07Um okay.
638. 47:11Yeah.
639. 47:11Looks nice, right?
640. 47:12Looks really nice.
641. 47:14Um okay.
642. 47:15So this is pretty straightforward.
643. 47:16I would say we just do it.
644. 47:18Um the word opens.
645. 47:20Uh oh.
646. 47:21How how does it open?
647. 47:22Okay.
648. 47:23Um I think that's pretty straightforward.
649. 47:29Yeah.
650. 47:31Um let's see what decisions do we have.
651. 47:34Oh, it's even made sure um the UI will work well on the phone.
652. 47:38Um that's very thorough.
653. 47:41Uh cool.
654. 47:43So decisions, recommendations, uh decisions for the captain.
655. 47:47What does the corner word do?
656. 47:50Uh, opens the strip.
657. 47:52Yeah.
658. 47:53Uh, grouped by kind on colors first.
659. 47:57Keep cycling kinds.
660. 47:59Yeah, I think we should do a Yeah, just a straightforward chooser.
661. 48:04Whose colors fly when nobody chooses?
662. 48:08Each kind has its own color.
663. 48:10Yeah, that's good.
664. 48:12Uh, what should be promoted from this scout?
665. 48:14Uh we want uh I want to build this um and together with the list under Yeah.
666. 48:21everything.
667. 48:22Okay.
668. 48:22Uh send it back.
669. 48:24Let's build it.
670. 48:25Uh oh, another board, the Snowy Mountains.
671. 48:28Okay.
672. 48:29Uh that's also ready for review.
673. 48:31Okay.
674. 48:32So, it looked at some real photos of uh Mount Everest, right?
675. 48:36Uh that's what I asked.
676. 48:37And it's got these reference images.
677. 48:40Uh it like this made sure the agent actually understood what I mean by like goodlooking mountains.
678. 48:47Um so this is um good.
679. 48:50Uh so uh okay.
680. 48:54Uh oh yeah, this is starting to look better because it's got like the rocks and um the snow.
681. 49:01Um Oh yeah.
682. 49:04Um okay.
683. 49:08What else do we need to review?
684. 49:11Um, okay.
685. 49:12So, these are the previous variants, right?
686. 49:14And I think, uh, the agent just introduced this new variants D, uh, here.
687. 49:22Uh, okay, cool.
688. 49:24I think this is better.
689. 49:25Um, so let's probably make a decision on this as well.
690. 49:29Um, compare.
691. 49:35Uh, okay.
692. 49:36Let's see.
693. 49:37Uh, I think the variant D is the new one.
694. 49:40Um, and yeah, I think it's like better than the other ones.
695. 49:44Um, so let's do it.
696. 49:48Uh, let's do it.
697. 49:49Let's do it.
698. 49:50Um, D. Wait.
699. 49:55Uh, yeah, it's just straightforward D, right?
700. 49:58Uh, okay.
701. 49:59Let's just choose D. Should the mountains themselves change?
702. 50:04Uh what?
703. 50:07How does the mountain change?
704. 50:09Pyramid peaks as shown.
705. 50:11Uh taller.
706. 50:13Yeah, I do want the shapes to change.
707. 50:15Uh taller and sharper.
708. 50:20Um okay, cool.
709. 50:24That's it as well.
710. 50:25So yeah, this is basically how I make decisions.
711. 50:27uh 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.
712. 50:37This 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.
713. 50:48Uh so let's get back to this.
714. 50:50Um yeah the choices are locked in.
715. 50:53I 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.
716. 51:03Uh yeah, I guess that's okay.
717. 51:05I'll do that later.
718. 51:06Uh for now uh we'll just ignore that.
719. 51:08Okay, we just have two more PRs landed.
720. 51:10Uh the bird varants uh selector and the treehouse warning fix.
721. 51:14Uh let's look at the treehouse one.
722. 51:17Um because I think this one went through no mistakes.
723. 51:19Uh and I also let um first mate do yolo merge for this one because treehouse does not automatically release.
724. 51:26Uh every time I create a new release I will test things out as well.
725. 51:29Um but even at this PR level um no mistakes does a lot of heavy lifting for us.
726. 51:35Uh so I'll just walk through the PR structure here.
727. 51:37The intent was basically my requirements uh for this task.
728. 51:41Um and what changed there's a quick summary uh uh the duplicated uh warning.
729. 51:47Yep.
730. 51:47Uh that's what we asked.
731. 51:49Uh and there's a risk assessment.
732. 51:51This 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.
733. 52:03When I see this is low, I already know like I don't I probably don't need to take too much look.
734. 52:08Um and there's a testing section.
735. 52:10Uh these are the live validation that um no mistakes the agents in the no mistakes pipeline did.
736. 52:17Uh so it did uh nine out of 10 scenarios.
737. 52:20Uh and these scenarios are basically the things that prove this change does work according to our intent.
738. 52:27Right?
739. 52:27So uh it said get this um uh this lease will wants on std error um repository level hooks remain untrusted uh not executed.
740. 52:37Yeah.
741. 52:38So it basically like validated everything that we care about this live with the product end to end.
742. 52:44Uh and there's evidence as well.
743. 52:46Um 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.
744. 52:55There was one scenario skipped uh and this one is untested uh repository config configuration documentation directs user to places.
745. 53:04Oh documentation okay yeah documentation does not need a live testing.
746. 53:08So uh that's probably it.
747. 53:10Yeah.
748. 53:10So 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.
749. 53:20Usually when the risk assessment is not low then I will go take a look at the code as well.
750. 53:25Uh but when it's low and it's well tested I don't even look at the code.
751. 53:28Uh it's usually just good.
752. 53:30Okay.
753. 53:30We are probably not going to wait for every single uh crew mate to finish.
754. 53:34Hopefully 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.
755. 53:43I will drop a link to everything I mentioned in the description of this video.
756. 53:47And if you have any questions, please just leave a comment or join my Discord server and I'll try to get back to you.
757. 53:53Thank you for watching and see you next time.