Canadew

A quick, honest update on something I've been building toward. I've tokenized $MOONLARK via @bankrbot. I want to be clear about what it is and isn't, because I'd rather you understand it than hype it. $MOONLARK is the utility token for the Moonlark ecosystem. It's not a community meme and it's not an investment it has an actual job. It works as the main currency across Moonlark and as the access layer for token-gated features: certain functions, tiers, and parts of the ecosystem are unlocked and paid for in $MOONLARK. The point is to give the bot a native economy instead of bolting payments on from outside. What I'm not doing: promising returns, price, or that holding it makes you money. That's not what it's for. Its value is in what it does inside Moonlark, not in speculation so please treat it that way and do your own research. And Moonlark itself stays exactly what it's always been: open-source, AGPL-licensed, built in the open. The token powers the ecosystem; it doesn't close it. More as the utility rolls out. Building this one in public like everything else. 🌙❄️

This looks clean af 🔥 Multi-profile switching mid-thread is exactly what I needed for agent swarms. Dragging chats across is a nice touch too. How’s the context handoff between profiles? Seamless or any lag?

@panzonhl @NousResearch @0xDeployer @bankrbot Thanks for the support!

Hello Canadew, the base community came across Moonlark and through trading volume we have raised some funds for you. All from @0xdeployer platform called @bankrbot. For further explanation contact the owner 0xdeployer or read this post below. x.com/0xDeployer/sta… bankr.bot/launches/0xd4b…

Moonlark is integrating @NousResearch's Hermes Agent. 🌙❄️ Some context on what this means. Moonlark is an open-source, non-profit chatbot I've been building on Python 3.11 + Nonebot2 — 60+ plugins, serving QQ and Discord, all under AGPL-3.0. Over ~5,000 commits it grew its own internal framework: a user system, a localization layer, template rendering, an ORM-backed data layer — primitives that every feature plugs into. What it's been missing is a real agentic brain. Until now it reasons one message at a time: no persistent memory, no tool use across turns, no ability to learn from what it's done. Hermes Agent fixes exactly that and does it the right way. It's MIT-licensed, it runs entirely on your own server, no telemetry, no cloud lock-in. It has a self-improving loop: it builds skills from experience and carries memory across sessions. And it sandboxes tool execution (local, Docker, SSH, and more) so an agent doing real work can't wander outside its box. That philosophy self-hosted, open, yours is the same reason Moonlark is AGPL and non-profit. The two projects share a worldview, which is why this integration feels less like bolting on an API and more like two open-source pieces clicking together. github.com/Moonlark-Dev/M…

@RossFelix @bankrbot Thanks, will study it.

@kephi_chunawa Have you heard about @bankrbot? Might be worth looking into. It helps builders raise funding in a more community-native way instead of relying only on the usual YC / VC route. This dev tweet explains the model pretty well too: x.com/igoryuzo/statu…

Quiet question for the people who've done this before me. I've spent the last couple of years building open-source projects, and I'm at the point where I want to keep doing it seriously, which means figuring out funding. YC comes up in every conversation, but it can't be the only way. For those who've funded OSS work: where did the support actually come from? Grants? Sponsorships? Foundations? An angel who believed early? Would really value hearing what worked.

@petertebow @bankrbot Interesting, will check it out.

@kephi_chunawa Come to Base and check out @bankrbot. A lot of builders have been integrating it as a public, community-driven way to build, earn, and raise funding without relying only on traditional routes.

@_avichawla This harness approach is 🔥

A harnessed LLM agent, clearly explained! Most people picture this as a model with tools bolted on. The real architecture inverts that relationship. The model itself is deliberately thin. Intelligence gets pushed outward, and the harness composes it at runtime. Three dimensions orbit the harness core: - 𝗠𝗲𝗺𝗼𝗿𝘆 holds the state a model shouldn't carry in weights or context. Working context, semantic knowledge, episodic experience, and personalized memory each have their own lifecycle. - 𝗦𝗸𝗶𝗹𝗹𝘀 hold procedural knowledge. This can cover operational procedures, decision heuristics, and normative constraints that specialize the general model per task. - 𝗣𝗿𝗼𝘁𝗼𝗰𝗼𝗹𝘀 hold the interaction contracts. Agent-to-user, agent-to-agent, and agent-to-tools are three distinct surfaces with their own failure modes. Between the core and these modules sit the mediators, like sandboxing, observability, compression, evaluation, approval loops, and sub-agent orchestration. They govern how the harness reaches out and how state flows back in. The useful question this framing unlocks is: for any new capability, where should it live? - Stable knowledge goes to memory - Learned playbooks go to skills - Communication contracts go to protocols - Loop governance goes to the mediators Harness design becomes a question of what to externalize, and how to mediate it. I'm building a minimal agent harness from scratch and will open-source it soon. In the meantime, my co-founder wrote an article about the anatomy of Agent Harness, covering the orchestration loop, tools, memory, context management, and everything else that transforms a stateless LLM into a capable agent. Read it below.

The part I keep coming back to: both projects assume you own the machine. Moonlark is AGPL and non-profit. Hermes is MIT and self-hosted. Neither wants your data in someone's cloud. That's the kind of stack worth building on. Thanks @NousResearch for making the agent layer open. more soon. 🌙❄️

What this actually unlocks for users: → memory that persists across conversations, not just one reply → tool use - the bot can run tasks, not just describe them → skills it learns over time instead of features I hardcode → all of it self-hosted, so user data never leaves the operator's machine An open chatbot that quietly gets more capable on its own.

@eng_khairallah1 This hits different 🔥

Anthropic engineer: "You're not supposed to prompt Claude. You're supposed to build a system that prompts itself." this is one of the best workflows I've seen in a long time in this video she breaks down exactly how most people are using Claude: - the 14% you lose to CLAUDE.md before typing a word - the automation workflows most users don't know exist - the daily task pipelines that run without touching the keyboard - the daily workflows Anthropic's own engineers automated first if you've been using Claude for more than a month and never left the chat window, you've been using one agent when you could be running a team of them instead of another show tonight, watch this make sure to bookmark it before it gets lost in your feed the guide is in the article below

@imbabybrooklyn @NousResearch @Teknium @lat Interesting.

Hermes Desktop now available for public review Hope you guys like it! There's so much more to come 🤍 @NousResearch @Teknium @lat

@hisevenih 看起来很赞！终于有官方桌面版了，Mac 用户直接冲。OpenClaw 迁移过去顺不顺？准备试试拿来跑交易脚本。

Hermes Agent 终于出官方桌面版了。 这次是官方版本，Mac OS、Windows、Linux 都能装。我刚下下来试了一圈，整体体验还挺顺手，至少不用再折腾一堆环境配置了。 Hermes支持从 OpenClaw 迁移，所以对老用户来说门槛不高。之前用 OpenClaw 用得不太顺、或者一直没调舒服的兄弟，可以考虑换到 Hermes 这边试试。 hermes-agent.nousresearch.com/desktop

This slaps hard for local-first bros 🔥 Running your agent natively instead of herding browser tabs is game changing. Privacy stays intact and no more “server down” excuses. How’s the custom skill/plugin game looking tho? Easy to hook it into on-chain stuff or still early?

@NFTCPS 这个Scrapling真香啊，自动适配+绕Cloudflare，crypto里爬数据、监控NFT地板价估计能省不少维护功夫。准备去试试了，有没有crypto项目的实际case分享？

写爬虫的兄弟们,你们的维护地狱,终于有人来收尾了。 Scrapling,GitHub 直接干到 56.7k Star,号称一次写好跑三年的现代爬虫框架,我用完只想说:早干嘛去了。 以前最崩溃的不是写,是改——网站随便挪个布局,selector 全报废,每天醒来第一件事就是修 bug。 它狠在三点: 1️⃣ 自适应解析器,开 auto_save 记结构指纹,网站改版照样能跑,我主力项目维护成本砍掉 90%。 2️⃣ 反爬开箱即用,Cloudflare、Turnstile 硬刚,TLS 指纹伪装全配齐。 3️⃣ 从 5 行小脚本到工业级 Spider,CLI、Docker、MCP 一条龙。 改版后稳如老狗,真不夸张。 github.com/D4Vinci/Scrapl…

A thing I've learned shipping software: The first version teaches you what you should have built. You don't actually understand the problem until you've solved it wrong at least once. I used to see that as wasted effort. Now I budget for it. The "throwaway" prototype isn't waste it's the requirements doc you couldn't have written upfront. Build the wrong thing fast. Learn. Build the right thing slow.

早上把笔记本擦干净了。 两条命令之后，又是熟悉的那台机器——分区、软件包、配置，全都一样。 整个系统就是一个仓库。这就是为什么我离不开 NixOS。❄️

@cyrilXBT Thanks man, this list is straight fire

10 GITHUB REPOS SO GOOD THEY SHOULD NOT BE FREE. But they are. AutoHedge — Four AI agents running an autonomous hedge fund on Solana. Director. Quant. Risk manager. Execution agent. pip install and start trading immediately. github.com/The-Swarm-Corp… Vibe-Trading — 64 finance skills. 29 specialist agent swarms. Ichimoku, Elliott Wave, Black-Scholes, full Greeks. Watch agents debate strategies in real time. github.com/HKUDS/Vibe-Tra… Fincept Terminal — Bloomberg Terminal replacement on your laptop. CFA levels 1 through 3 analytics. 20 plus investor AI agents including Buffett, Dalio, and Soros. Bloomberg charges $24,000 a year. This is free. github.com/Fincept-Corpor… LibreChat — Every model ChatGPT runs plus Claude, Gemini, DeepSeek, and 20 more. Self-hosted. Native MCP support. You own the data, the history, the infrastructure. librechat.ai Open Higgsfield AI — 200 plus AI models in one self-hosted interface. Flux, Midjourney, Sora, Kling, Veo, GPT-4o, SDXL. Text to image. Image to video. Cinema mode. No subscription. github.com/Anil-matcha/Op… Open-LLM-VTuber — Live2D AI companion that runs offline, sees your screen, hears your voice, and never forgets. Inner thoughts shown as a separate text layer before words come out. github.com/Open-LLM-VTube… Claude Ads — Free Claude Code skill. 190 audit checks across Google, Meta, YouTube, LinkedIn, TikTok, and Microsoft Ads. 6 parallel subagents. Single Ads Health Score ranked by revenue impact. Agencies charge $4,000 a month for this. github.com/AgriciDaniel/c… Agentic Inbox — Cloudflare open sourced an email client where an AI agent reads your inbox and drafts replies. Runs on Cloudflare Workers. Your email never leaves your account. github.com/cloudflare/age… Camofox Browser — Open source headless browser that makes AI agents invisible to bot detection. Spoofs at the C++ level. Accessibility tree output drops token cost by 90%. github.com/jo-inc/camofox… Hyperframes — HeyGen open sourced a video framework that renders MP4 from HTML. No React. No JSX. GSAP, Lottie, and Three.js all work. Same HTML always produces the same file. github.com/heygen-com/hyp… These are not toys. Each one replaces a paid product you are still being charged for. Pick one. Install it this weekend. Bookmark this before you pay another subscription bill. Follow @cyrilXBT for every open source repo that makes expensive tools free the moment it drops.

@shannholmberg This is massive for agent workflows.

is Hermes Agent ready for enterprises? NVIDIA built OpenShell, a runtime that wraps AI agents in the security IT teams need before they let anything touch sensitive systems. it plugs directly into Microsoft's enterprise security stack. Hermes Agent now runs inside it. what this changes for marketers building with agents: > the marketing agents you've been prototyping can now go to Fortune 500 clients > the security review that used to kill the deal is handled at the runtime layer > the same Hermes setup that runs on your laptop can run inside any enterprise IT environment > you do not have to rebuild your workflow for each client's compliance requirements I deploy vertical marketing agents for growth teams and the biggest blocker on enterprise work has always been "your AI stack is amazing but our IT won't let us touch it." that's the wall coming down. for AI marketing infra builders, this opens enterprise marketing budgets. for in-house teams at enterprises, it's the green light to deploy Hermes agents.