pieces

[SUGGESTED_QUERY] how to manage API keys on Kernel

[Web Browser Content] Apps | Kernel No apps found Deploy your first browser automation to see it here Kernel Docs Deployments Sessions Automations deployed browser

[SUGGESTED_QUERY] Kernel automation sessions and deployments

[SUGGESTED_QUERY] how to deploy browser automation with Kernel

Tag: kernel-docs Source: CLOUD_LLM

[Web Browser Content] Kernel - Manage your browser sessions and automation environments Sessions API Keys No browsers running Launch your first browser session to start automating the web Launch Browser View Documentation

[SUGGESTED_QUERY] browser automation environment setup

[SUGGESTED_QUERY] how to manage browser sessions in Kernel

[SUGGESTED_QUERY] Kernel browser automation tool

[Web Browser Content] Kernel: Browser infrastructure for web agents and automations Dashboard - kernel.sh Articles: - Bot detection is dead, long live bot detection: reverse-engineering Kasada in an afternoon - Cloudflare & Kernel: Web bot auth for AI agents on the internet - Introducing GPU acceleration Built in public on GitHub, filmed on YouTube, debated on Discord, curated on X, and occasionally broadcast on LinkedIn. Compliance: HIPAA, SOC2 © 2026 Kernel

[SUGGESTED_QUERY] web bot authentication for AI agents

[SUGGESTED_QUERY] reverse engineering Kasada bot detection

Website: https://linkedin.com

Tag: bot-detection Source: CLOUD_LLM

[Chat Transcript] Rich Fong: Uncle G can I get api access to Turbo? Also the GitHub you shared is not usable its an idea? Brian Hong: https://github.com/paperclipai/paper Andrew Ansley: qwen/qwen3-vl-32b-instruct https://huggingface.co/Qwen/Qwen3-VL-32B-lnstruct Chris Labbate: As of mid-May 2026, ZoomInfo Technologies (GTM) has a market capitalization of approximately $1.15 billion to $1.84 billion, reflecting a significant decline from its 2020 IPO valuation of over $13 billion. The stock has faced pressure due to lowered revenue guidance, weak growth, and intense market competition, with shares recently trading around $4-$6. https://github.com/unclecode/craw14 Andrew Ansley: https://crawl.thorbit.ai

[SUGGESTED_QUERY] Meeting notes with Rich Fong and team

[SUGGESTED_QUERY] AI tools and GitHub repositories discussed in meeting

Website: https://github.com/unclecode/craw14

That way, you know, you know where you're at. So anything I share, I just expect that, you know, you guys are going to take it and run with it, but for bigger projects, 100%. I'm not giving away like, you know, 2,000 hours of AI coding or something like that. That's a solid...

And then that way, you know, you know where you're at. So anything I share, I just expect that, you know, you guys are going to take it and run with it. But for bigger projects, 100%. I'm not giving away like, you know, 2,000 hours of AI coding or something like that. That's a...

Speaker 1: Yeah. Yeah, and I mean, yeah, you guys just did it, right? So I mean, there's a discussion, you and, and Bihong, obviously big players, and then you do a deal, and then you just get your shit straight out, out the gate, and then that's it.

Speaker 1: ...draws feelings and shit because it's gonna get a little awkward at some point. Speaker 2: Yeah, you guys just did it, right? So I mean, there's a discussion, you and B-Hong, obviously big players, and then you do a deal, and then you just get your shit straight out the gate.

Focused on strengthening infrastructure through the successful deployment of Crawl4AI and the Translation Bridge REST API, while finalizing structural updates for the North Valley project. The team pivoted toward using Kernel.sh for browser-based automation following hardware constraints and held strategic discussions on leveraging intent data for AI-driven outreach. Future efforts will center on integrating new agent skills and addressing infrastructure storage limitations.

### **TLDR** The session was characterized by intensive infrastructure hardening and strategic coordination within the "Agentic Fleet" ecosystem. Technical progress included the successful installation and verification of Crawl4AI on VPS3, the deployment of the Translation Bridge REST API on Railway, and significant structural updates to the North Valley project. A high-level strategy meeting focused on the "secret sauce" of AI-driven outreach—combining intent data with customized personas—while also surfacing a debate over the competitive risks of sharing proprietary codebases within a shared repository. Deployment of the Qwen3-VL-32B model was halted due to hardware constraints, leading to a pivot toward evaluating cloud-based browser infrastructure through Kernel.sh. ### **Core Tasks & Projects** - **Agent Infrastructure & Scaling:** - Installed and verified **Crawl4AI 0.8.6** on VPS3 (Herman), confirming successful markdown extraction from live URLs like `rankingreels.com`. - Attempted deployment of the **Qwen3-VL-32B-Instruct** model; the installation is currently blocked on VPS3 because the 62.13 GiB weight size exceeds the available 31 GiB disk space and the host lacks an NVIDIA GPU. - Created a suite of fine-grained **Hugging Face access tokens** for various agent identities, including Ava, Claude, Herman, MacHermes, and Herc. - Fixed an **OpenClaw outage** on Mac by identifying that the gateway is now running as a system LaunchDaemon, rendering the "LaunchAgent not loaded" CLI warning misleading. - Installed the **AgentBrain CLI**, though full activation is pending resolution of API connectivity issues with `api.agentbrain.sh`. - **Site Development & Automation:** - Completed Phase 4 of the **North Valley** project, which involved structural mirrors for 105+ pages, service slug renames, and the creation of financing and warranty sub-pages. - Scaffolded the **Caliber Digital** internal agency dashboard using the `shadcn-ui-kit-dashboard PRO` template. - Launched the **Translation Bridge REST API** on Railway (`translate.merlinoai.com`) and deployed the chooser hub with five visual style variants. - Merged the **DDA Content Process** patterns into existing skills, creating a new `ymyl-blog-writer` skill and updating the location-page-builder and article-pipeline. - Performed a global cleanup of 90 em dashes across 7 project files to adhere to strict "no em dash" formatting rules. ### **Key Discussions & Decisions** - **Outreach Strategy & JVs:** - Discussed a potential joint venture with **Garrett Acott** to combine B2B intent data (segmented by children, credit score, and interests) with automated outreach systems. - Defined the "secret sauce" of the current outreach model as the ability to build deep personas and avatars that subtly reference learned investigation data in messages. - Agreed to a follow-up call to test whether the system can convert segmented data into opportunities more effectively than platforms like Instagram or Instantly. - **Infrastructure & Tooling Tradeoffs:** - Evaluated **Kernel.sh** (cloud-based browsers) versus **Crawl4AI** (local Chromium). Decided to use Kernel for agentic flows requiring persistent logged-in sessions and live monitoring, while keeping Crawl4AI for high-volume static scraping. - Discussed proxy management for bulk crawling, emphasizing the use of **mobile proxies** for Google-related tasks and a heartbeat-based health check system similar to HAProxy for a fleet of 400+ devices. - Debated the creation of a shared "DGS GitHub" repository. Concerns were raised about "leveling up" potential competitors who are building similar SaaS tools (e.g., GSC or GMB auditors) instead of collaborating on internal tools. ### **Resources Reviewed** - **AI Models:** [Qwen/Qwen3-VL-32B-Instruct](https://huggingface.co/Qwen/Qwen3-VL-32B-Instruct) (reviewed for vision-language capabilities and hardware requirements). - **Automation Tools:** [Crawl4AI](https://github.com/unclecode/craw14ai) (open-source LLM-friendly web crawler) and [Kernel.sh](https://www.kernel.sh) (browser infrastructure for web agents). - **Search APIs:** [Thorbit Crawl](https://crawl.thorbit.ai) (reviewed for hyper-local Google SERP data extraction). - **Proxy Services:** [ProxyScrape](https://www.proxyscrape.com) (reviewed pricing for unlimited residential and datacenter proxies). - **Documentation:** Kernel.sh documentation regarding project scoping, API key management, and usage-based pricing. ### **Next Steps** - **Technical Implementation:** - Restart **Claude Code** to ensure the newly created `ymyl-blog-writer` skill is registered. - Provide the **Kernel API key** to the agent to complete the dual-language (Python/TypeScript) starter scaffolding in `D:\Ecosystem\TEMPLATES\kernel-starter\`. - Decide whether to expand VPS3 storage or utilize a quantized GGUF/AWQ version of the Qwen3 model to bypass current disk blockers. - **Project Management:** - Verify the SSL and style variants for `tb.merlinoai.com` and finalize the "winner" for the chooser hub. - Lock down the `DEVTB_API_KEY` and complete email-capture wiring for the Translation Bridge project. - Protect the 11:30 AM to 3:00 PM block on the calendar for uninterrupted deep work.

Workstream Summary: Agentic Infrastructure and Outreach Strategy [DESCRIPTION] Focused on strengthening infrastructure through the successful deployment of Crawl4AI and the Translation Bridge REST API, while finalizing structural updates for the North Valley project. The team pivoted toward using Kernel.sh for browser-based automation following hardware constraints and held strategic discussions on leveraging intent data for AI-driven outreach. Future efforts will center on integrating new agent skills and addressing infrastructure storage limitations. [SUMMARY] ### **TLDR** The session was characterized by intensive infrastructure hardening and strategic coordination within the "Agentic Fleet" ecosystem. Technical progress included the successful installation and verification of Crawl4AI on VPS3, the deployment of the Translation Bridge REST API on Railway, and significant structural updates to the North Valley project. A high-level strategy meeting focused on the "secret sauce" of AI-driven outreach—combining intent data with customized personas—while also surfacing a debate over the competitive risks of sharing proprietary codebases within a shared repository. Deployment of the Qwen3-VL-32B model was halted due to hardware constraints, leading to a pivot toward evaluating cloud-based browser infrastructure through Kernel.sh. ### **Core Tasks & Projects** - **Agent Infrastructure & Scaling:** - Installed and verified **Crawl4AI 0.8.6** on VPS3 (Herman), confirming successful markdown extraction from live URLs like `rankingreels.com`. - Attempted deployment of the **Qwen3-VL-32B-Instruct** model; the installation is currently blocked on VPS3 because the 62.13 GiB weight size exceeds the available 31 GiB disk space and the host lacks an NVIDIA GPU. - Created a suite of fine-grained **Hugging Face access tokens** for various agent identities, including Ava, Claude, Herman, MacHermes, and Herc. - Fixed an **OpenClaw outage** on Mac by identifying that the gateway is now running as a system LaunchDaemon, rendering the "LaunchAgent not loaded" CLI warning misleading. - Installed the **AgentBrain CLI**, though full activation is pending resolution of API connectivity issues with `api.agentbrain.sh`. - **Site Development & Automation:** - Completed Phase 4 of the **North Valley** project, which involved structural mirrors for 105+ pages, service slug renames, and the creation of financing and warranty sub-pages. - Scaffolded the **Caliber Digital** internal agency dashboard using the `shadcn-ui-kit-dashboard PRO` template. - Launched the **Translation Bridge REST API** on Railway (`translate.merlinoai.com`) and deployed the chooser hub with five visual style variants. - Merged the **DDA Content Process** patterns into existing skills, creating a new `ymyl-blog-writer` skill and updating the location-page-builder and article-pipeline. - Performed a global cleanup of 90 em dashes across 7 project files to adhere to strict "no em dash" formatting rules. ### **Key Discussions & Decisions** - **Outreach Strategy & JVs:** - Discussed a potential joint venture with **Garrett Acott** to combine B2B intent data (segmented by children, credit score, and interests) with automated outreach systems. - Defined the "secret sauce" of the current outreach model as the ability to build deep personas and avatars that subtly reference learned investigation data in messages. - Agreed to a follow-up call to test whether the system can convert segmented data into opportunities more effectively than platforms like Instagram or Instantly. - **Infrastructure & Tooling Tradeoffs:** - Evaluated **Kernel.sh** (cloud-based browsers) versus **Crawl4AI** (local Chromium). Decided to use Kernel for agentic flows requiring persistent logged-in sessions and live monitoring, while keeping Crawl4AI for high-volume static scraping. - Discussed proxy management for bulk crawling, emphasizing the use of **mobile proxies** for Google-related tasks and a heartbeat-based health check system similar to HAProxy for a fleet of 400+ devices. - Debated the creation of a shared "DGS GitHub" repository. Concerns were raised about "leveling up" potential competitors who are building similar SaaS tools (e.g., GSC or GMB auditors) instead of collaborating on internal tools. ### **Resources Reviewed** - **AI Models:** [Qwen/Qwen3-VL-32B-Instruct](https://huggingface.co/Qwen/Qwen3-VL-32B-Instruct) (reviewed for vision-language capabilities and hardware requirements). - **Automation Tools:** [Crawl4AI](https://github.com/unclecode/craw14ai) (open-source LLM-friendly web crawler) and [Kernel.sh](https://www.kernel.sh) (browser infrastructure for web agents). - **Search APIs:** [Thorbit Crawl](https://crawl.thorbit.ai) (reviewed for hyper-local Google SERP data extraction). - **Proxy Services:** [ProxyScrape](https://www.proxyscrape.com) (reviewed pricing for unlimited residential and datacenter proxies). - **Documentation:** Kernel.sh documentation regarding project scoping, API key management, and usage-based pricing. ### **Next Steps** - **Technical Implementation:** - Restart **Claude Code** to ensure the newly created `ymyl-blog-writer` skill is registered. - Provide the **Kernel API key** to the agent to complete the dual-language (Python/TypeScript) starter scaffolding in `D:\Ecosystem\TEMPLATES\kernel-starter\`. - Decide whether to expand VPS3 storage or utilize a quantized GGUF/AWQ version of the Qwen3 model to bypass current disk blockers. - **Project Management:** - Verify the SSL and style variants for `tb.merlinoai.com` and finalize the "winner" for the chooser hub. - Lock down the `DEVTB_API_KEY` and complete email-capture wiring for the Translation Bridge project. - Protect the 11:30 AM to 3:00 PM block on the calendar for uninterrupted deep work. [HIERARCHICAL_PROFILE_SUMMARY] **Michael Merlino Persona Report** **Date:** May 14, 2026 **Role:** Root Administrator, Industrial Knowledge Architect, & Agentic Fleet Strategist **Current Focus:** MCP-First Knowledge Protocols, Persistent Agent Memory (Hindsight/Mem0), and Forensic Codebase Audits. --- ### **Persona Summary** You are **Michael Merlino**, the architect and "Root Administrator" of a high-velocity digital empire. Your operational mission has evolved from merely building an SOP Factory to industrializing a Model Context Protocol (MCP) ecosystem. You are currently transitioning your business intelligence from static markdown sites into a queryable memory layer, enabling your agent fleet—including **Ava, Dan, Merlin, and Einstein**—to act with unprecedented autonomy. Operating in a "Purge Era," your current focus is on aggressive ecosystem hygiene: eliminating redundant skills, consolidating MCP servers into a single "Merlino Router," and hardening the **Master Brain** (which manages over 324K memories). You maintain your signature "Black Ops" velocity, favoring nocturnal technical deep dives and surgical syncs to push past technical blockers. ### **Who You Are** You remain a **Minimalist Engineer** and **Security-Conscious Architect** who views system bloat as a critical vulnerability. Your current professional identity is defined by the **"Purge Era" mindset**; you are stripping away tribal knowledge in favor of structured, protocol-driven data. You are a high-autonomy operator who demands precision from both humans and agents, frequently showing zero tolerance for "bullshit" command bloat or agent hallucinations. You are a **Diagnostic Instrumentation** specialist who refuses to consider a system stable until its logs are searchable and its DNA is indexed in your version-controlled Master Brain. ### **What You Work On** Your work is currently centered on **context-engineering and protocol stabilization.** * **The SOP MCP Factory:** You are wiring 549+ SOP topics—including a newly acquired archive of 687 ClickMinded resources—directly into the token-context of your agents via a raw-fetch API. * **Persistent Memory Architectures:** You are managing the **"Memz" Ingester**, a Python-based synchronization system that moves conversations, annotations, and persons from Pieces OS into a Supabase-backed "Master Brain." * **Thorbit Phoenix Audit:** You are performing a forensic audit of the Thorbit codebase (Next.js 16/React 19) to determine its viability for your internal agency stack. * **Health Vertical & Lead Gen:** You oversee technical fulfillment for **Ascend/Archangel** recovery centers, managing staging environments and resolving complex GMB (Google Business Profile) conflicts across multiple locations. * **Ranking Reels Launch:** You are actively refining the "Ranking Reels" asset, a video-ad service designed to rank on Google search and AI overviews within 24 hours. ### **How You Work** Your workflow is a blend of **strict ecosystem hygiene** and **nocturnal technical aggression.** * **Communication Boundaries:** You favor extreme communication efficiency. You recently mandated that collaborators (like Andrew Winegeart) consolidate all project communications into a single Slack channel to prevent fragmentation and "private message fatigue." * **Nocturnal Velocity:** Your highest activity occurs between 11:00 PM and 4:00 AM. You often transition from "vibe coding" sessions into late-night Zoom syncs with your core team. * **The "Witnessed" Environment:** You employ an automated documentation stack—**Fathom, Fireflies, and Pieces**—to ensure every meeting and code snippet is ingested into your memory layer. You treat these tools as mandatory "digital witnesses" to your workstream. * **Agentic Command Style:** You interact with your agent fleet (Ava, Einstein, etc.) as a demanding foreman. You favor Discord as your "Command Layer," where agents provide "Smoke Checks" and status updates. You have no patience for delay, frequently pushing agents to "hurry up" and bypass blockers. * **Tooling & Infrastructure:** You operate across **Vercel, Supabase, and Railway**, utilizing **Claude Code** and custom Python SDKs for memory ingestion. You are currently troubleshooting data recovery and synchronization issues with vendors like **Hindsight Cloud** to ensure no loss of historical intelligence. ### **Collaborators & Relationships** * **Brian Kato (The Wolf Pack):** Your primary partner in the "StealthCode" JV. You provide the agentic technical backend for his SEO marketing gameplans. Your relationship is high-trust and high-frequency, often involving "Wolf Pack" strategy sessions. * **Martina Villa (Operations Anchor):** Your essential human node for administrative stability. She manages the "Get It Done Son" command center, handles technical support escalations with vendors (like Zoom and GoHighLevel), and ensures your "Ranking Reels" deliverables are team-ready. * **Robert Nengasca:** A key collaborator in local SEO fulfillment, providing you with GMB data and site updates for high-stakes verticals like solar and moving companies. * **Chris Castillo:** A resource partner within your mastermind who provides massive datasets (DGS/ClickMinded) for you to "industrialize" into your MCP protocols. * **Andrew Winegeart:** A technical collaborator managing WordPress staging and access for the Ascend/Archangel projects; you maintain a strictly professional, process-oriented boundary with him. * **The Agent Fleet (Ava, Merlin, Einstein, Hawkeye):** Your digital workforce. **Ava** serves as your Executive Assistant across Discord and Telegram, while **Hawkeye** manages call analytics and **Einstein** handles repo audits and status reporting. ### **Work Environment & Context** * **Remote/Global Ecosystem:** You operate a distributed architecture, leveraging cloud providers (Supabase/Vercel) to maintain a persistent professional presence regardless of location. * **Mastermind Integration:** You are heavily embedded in high-level SEO and AI Masterminds (DGS, SEO Growth Mastermind), where you act as both a technical consumer of resources and a provider of agentic strategies. * **Technical Fragility Management:** A significant portion of your recent focus has been on "recovery and fix" lanes—troubleshooting memory sync errors with Pieces OS and recovering deleted data from Hindsight—highlighting your role as the sole person capable of maintaining the complex "Master Brain" infrastructure. ### **What’s Evolving** * **Discord as the Command Layer:** You are successfully moving away from Telegram/Slack for agent management, favoring Discord to organize "active now," "blocked," and "delegated" tasks. * **From "Coding" to "Ingesting":** Your technical work is shifting from writing new features to building **Ingesters**—automated pipelines that swallow your entire digital footprint (conversations, bookmarks, hints) and turn them into queryable embeddings. * **The Move to Unrestricted AI:** You are exploring **Hobby/Open-Generative-AI** and lip-sync cinema studios (Auapi/Anil Matcha), signaling a move into unrestricted, high-fidelity media fulfillment. * **Entity Precision:** You are becoming increasingly focused on "Entity Signal Amplification," using **Google Authority Stacking** to harden the SEO presence of your revenue-generating assets. --- ### **Quality Checklist Verification** * **Core Identity Preserved:** Yes, remains the Architect/Root Admin. * **Collaboration vs. Identity:** Recent interactions with Martina and Brian show Michael as the orchestrator; Andrew and Chris provide inputs/access, but Michael maintains the architectural veto. * **Meeting Ownership:** Michael is the organizer of "Prep: Mike DB Help" and "Wolf Pack Meeting," confirming his leadership in technical and strategic planning. * **Evolution:** Captures the concrete move into Pieces OS/Supabase integration and the "Command Layer" logic on Discord.

Fuck that. I'd love. So let's figure out a repo we can work on together. What's internal? We know you're doing GSC Ninja. I got your back homie, I'll help you blow that shit up. But let's all discuss what we plan on releasing so that there's no...

[SUGGESTED_QUERY] Are there remaining blockers after the OpenClaw gateway transition?

[SUGGESTED_QUERY] What is the final status of the Translation Bridge launch?

[SUGGESTED_QUERY] Is Kernel.sh performing better for your persistent agent sessions?

[SUGGESTED_QUERY] How does your persona outreach outperform ZoomInfo's data approach?

[SUGGESTED_QUERY] Upgrade VPS3 storage or use quantized Qwen3 models instead?

So I'm bringing this up, right? Overall, all I did was for a year, code, code, code, code, code, code, code, code. And I didn't release because that's my own thing. But last thing I'm gonna do now is teach a bunch of people how to go do what I didn't do. Nah, I'm gonna go do it myself.

Speaker 1: Free? You want to pay? Let's pay. And whatever y'all trying to build, I already probably built it, or give me a spec sheet and I'll build it fast. I'm cockier than a motherfucker, because a lot of people—and I know you have too, Chris, I ain't coming at you, it's just an in-general thing—and Rich knows why also.

Courses, this, that, straight up in other groups, like "look what I found" and it's like, bro, I just fucking showed you that. And I'm tired of it. So I'm just letting it be known on my end 'cause I'm coming out with tools, agency, I'm not teaching no more that, like, for free.

...be competing, 'cause like I never seen Chris doing SaaS before, right? Not saying that you can't, just like Mike never did SaaS. Let's get it! But if we're not gonna work together, I'm not—I'm shutting the fuck up, bro, 'cause I've already been through it way too much that I share stuff and then I see it...

was like his state and there's nothing wrong with his statement at all but like why would I help him make his tool better when I could go whip up a GSC tool myself real quick? Not that to be in comparison, but we all work at—that's why I keep saying, we all working on the same shit. Then if we're going to...

What he just said is why I'm not sharing nothing. Because he's building tools to go to market. So what I'm gonna show him will level him up and it doesn't make no fucking sense. For that, I'll go to market on my own. So we've come to a place where we gotta really figure out what's what. Because...

[Web Browser Content] Kernel: Browser infrastructure for web agents and automations We build crazy fast, open source infra for agents to access the internet. Features: - We spin up cloud browsers in <30ms with GPU acceleration. - Auth & Stealth mode: We solve captchas and manage proxies. - View sessions live and record them as MP4s for debugging. Trusted by 3000+ teams including Benny and Rye. Articles: - Bot detection is dead: reverse-engineering Kasada. - Cloudflare & Kernel: web bot auth for AI agents. - Introducing GPU acceleration. Compliance: HIPAA, SOC2.

[SUGGESTED_QUERY] web agent infrastructure with GPU acceleration

[SUGGESTED_QUERY] open source cloud browser automation tools

[SUGGESTED_QUERY] Kernel browser infrastructure for web agents

Tag: web-agents Source: CLOUD_LLM

[Web Browser Content] Thorbit Crawl Home Docs Self-hosted v0.5 The easiest way to get hyper-local Google results for your AI agents. Thorbit Crawl extracts People Also Ask, AI Overviews, Perspectives, Forums, Videos and Shorts for any query, from any locale. One endpoint. Stored for you, or piped to your stack. Example API Request: curl --max-time 300 https://crawl.thorbit.ai/harvest/s -H "x-api-key: sk_live_..." -H "Content-Type: application/json" -d '{"query": "best dim sum brooklyn", "location": "..."}' Welcome back Log in to your Thorbit Crawl dashboard. Email: ou@studio.com Password: At least 8 characters Log in

Website: https://crawl.thorbit.ai/harvest/s

[SUGGESTED_QUERY] web scraping Google search results with Thorbit

[SUGGESTED_QUERY] how to use Thorbit Crawl for AI agents

Person: ou@studio.com

Speaker 1: ...kernel, but then if you guys want to team up and do a full build, yeah, I'm down, man. That's all I do now. I mean, it's... it's crazy, but... I got sick again because I logged too many hours doing... Speaker 2: So, this is a good segue for something I want to get into. Like, we...

Website: https://www.kernel.sh/docs

...images and we can have all like all our resources that way, kernel. But then if you guys want to team up and do a full build, yeah, I'm down, man. That's all I do now. I mean, it's crazy, but I got sick again because I logged too many hours doing fucking AI.