Loading video...

Video Failed to Load

There was a problem loading this video. This could be due to a temporary network issue or the video might be unavailable.

Introducing React Doctor Scan your React codebase for anti-patterns: - Unnecessary useEffects - Fix accessibility issues - Prop drilling instead of context / composition Run as a CLI or agent skill. Repeat until passing. Fully open source

Aiden Bai

69,587 subscribers

1,077,618 views • 5 months ago •via X (Twitter)

Science & Technology

Anya Rossi• Live Now

Private livecam show

0 Comments

No comments available

Comments from the original post will appear here

Related Videos

Introducing /react-doctor Your React app probably has bad code. This fixes it Install as agent skill. Fully open source. npx react-doctor@latest

Introducing /react-doctor Your React app probably has bad code. This fixes it Install as agent skill. Fully open source. npx react-doctor@latest

Aiden Bai

580,833 views • 1 month ago

Introducing Expect Let agents test your code in a real browser 1. Run Claude Code / Codex to QA your app 2. Watch a video of every bug found 3. Fix and repeat until passing Run as a CLI or agent skill. Fully open source

Introducing Expect Let agents test your code in a real browser 1. Run Claude Code / Codex to QA your app 2. Watch a video of every bug found 3. Fix and repeat until passing Run as a CLI or agent skill. Fully open source

Aiden Bai

842,510 views • 3 months ago

Introducing React Native Doctor Find performance and security bugs in your mobile app Run via CLI and fix with agents. Fully open source npx react-doctor@latest

Introducing React Native Doctor Find performance and security bugs in your mobile app Run via CLI and fix with agents. Fully open source npx react-doctor@latest

Aiden Bai

170,826 views • 1 month ago

Introducing Amicro. A curated library of premium micro-interactions and transition components for React. Add beautiful card hovers, button interactions, and seamless transitions to your app with a single CLI command. ✨ Open source. Try it:

Introducing Amicro. A curated library of premium micro-interactions and transition components for React. Add beautiful card hovers, button interactions, and seamless transitions to your app with a single CLI command. ✨ Open source. Try it:

Kiyotaka

366,157 views • 4 days ago

🚨 NOW YOU RUN A COMPANY WITH ZERO EMPLOYEES Paperclip is a 100% open-source framework (70k+ stars) that makes this possible. Rather than just prompting a model, you hire a CEO, engineers, and a QA reviewer. Every worker is an AI agent, and Paperclip is the Node.js and React control plane that keeps them aligned. Stop chaining messy scripts together and build a living organization: → Stand up a CEO agent to set strategy → Hire engineers and designers via Claude or Codex → Build in an automated QA loop before any ticket closes → Manage the entire portfolio from your phone When an agent slips, you do not rewrite your whole pipeline: you just correct its persona prompt, exactly like coaching a junior hire. It is exactly the kind of tooling the space needs right now. Free, open-source, and self-hosted. Repo link in 🧵↓

🚨 NOW YOU RUN A COMPANY WITH ZERO EMPLOYEES Paperclip is a 100% open-source framework (70k+ stars) that makes this possible. Rather than just prompting a model, you hire a CEO, engineers, and a QA reviewer. Every worker is an AI agent, and Paperclip is the Node.js and React control plane that keeps them aligned. Stop chaining messy scripts together and build a living organization: → Stand up a CEO agent to set strategy → Hire engineers and designers via Claude or Codex → Build in an automated QA loop before any ticket closes → Manage the entire portfolio from your phone When an agent slips, you do not rewrite your whole pipeline: you just correct its persona prompt, exactly like coaching a junior hire. It is exactly the kind of tooling the space needs right now. Free, open-source, and self-hosted. Repo link in 🧵↓

Charly Wargnier

37,028 views • 23 days ago

React Native now has its own shadcn/ui equivalent — introducing 𝗡𝗮𝘁𝗶𝘃𝗲𝗨𝗜. If you love the flexibility of copying customisable components directly into your project (avoiding heavy, dependency-laden packages), NativeUI is designed for you. 𝗡𝗮𝘁𝗶𝘃𝗲𝗨𝗜 offers beautifully crafted, accessible components tailored for React Native, following the same copy-paste philosophy as shadcn/ui. Built with 𝗡𝗮𝘁𝗶𝘃𝗲𝗪𝗶𝗻𝗱 for fast, declarative, and flexible styling optimised for React Native. ➡️ 𝗖𝗼𝗽𝘆 𝗰𝗼𝗺𝗽𝗼𝗻𝗲𝗻𝘁 𝗰𝗼𝗱𝗲 𝗱𝗶𝗿𝗲𝗰𝘁𝗹𝘆 𝗶𝗻𝘁𝗼 𝘆𝗼𝘂𝗿 𝗽𝗿𝗼𝗷𝗲𝗰𝘁 — no black-box dependencies required. ➡️ 𝗖𝗼𝗺𝗽𝗼𝗻𝗲𝗻𝘁𝘀 𝗮𝗿𝗲 𝗮𝗰𝗰𝗲𝘀𝘀𝗶𝗯𝗹𝗲 𝗯𝘆 𝗱𝗲𝗳𝗮𝘂𝗹𝘁, supporting screen readers and keyboard navigation, and designed to align with native iOS and Android UX patterns. ➡️ 𝗙𝘂𝗹𝗹 𝗰𝗼𝗻𝘁𝗿𝗼𝗹 𝗼𝘃𝗲𝗿 𝘆𝗼𝘂𝗿 𝗨𝗜 without rebuilding common elements like buttons, inputs, or sliders from scratch. ➡️ 𝗖𝗼𝗺𝗽𝗮𝘁𝗶𝗯𝗹𝗲 𝘄𝗶𝘁𝗵 𝗘𝘅𝗽𝗼 𝗮𝗻𝗱 𝘃𝗮𝗻𝗶𝗹𝗹𝗮 𝗥𝗲𝗮𝗰𝘁 𝗡𝗮𝘁𝗶𝘃𝗲 𝗽𝗿𝗼𝗷𝗲𝗰𝘁𝘀, but not yet integrated with Tamagui’s styling system (future support may be planned). ➡️ 𝗦𝘂𝗽𝗽𝗼𝗿𝘁𝘀 𝘁𝗵𝗲𝗺𝗶𝗻𝗴 𝘃𝗶𝗮 𝗡𝗮𝘁𝗶𝘃𝗲𝗪𝗶𝗻𝗱 — though you’ll need to wire it up manually using Tailwind variables, context providers, and config files. Note: The term “install” in the documentation refers to using the shadcn CLI (e.g., npx shadcn@latest add component) to fetch and copy component code into your project, not adding a package to your dependencies. NativeUI isn’t a plug-and-play library; it’s a lightweight toolbox that empowers you to shape your UI with precision and control. 𝗪𝗵𝗮𝘁’𝘀 𝘆𝗼𝘂𝗿 𝗽𝗿𝗲𝗳𝗲𝗿𝗲𝗻𝗰𝗲: npm install a pre-built UI kit for speed, or copy/paste NativeUI components for ultimate customisation? #ReactNative #KeyboardUX #MobileDev #OpenSource #JSDev #Performance #iOSDev #KeyboardExtensions #ReactNativeKeyboard #UIUX #shadcn #nativeui

React Native now has its own shadcn/ui equivalent — introducing 𝗡𝗮𝘁𝗶𝘃𝗲𝗨𝗜. If you love the flexibility of copying customisable components directly into your project (avoiding heavy, dependency-laden packages), NativeUI is designed for you. 𝗡𝗮𝘁𝗶𝘃𝗲𝗨𝗜 offers beautifully crafted, accessible components tailored for React Native, following the same copy-paste philosophy as shadcn/ui. Built with 𝗡𝗮𝘁𝗶𝘃𝗲𝗪𝗶𝗻𝗱 for fast, declarative, and flexible styling optimised for React Native. ➡️ 𝗖𝗼𝗽𝘆 𝗰𝗼𝗺𝗽𝗼𝗻𝗲𝗻𝘁 𝗰𝗼𝗱𝗲 𝗱𝗶𝗿𝗲𝗰𝘁𝗹𝘆 𝗶𝗻𝘁𝗼 𝘆𝗼𝘂𝗿 𝗽𝗿𝗼𝗷𝗲𝗰𝘁 — no black-box dependencies required. ➡️ 𝗖𝗼𝗺𝗽𝗼𝗻𝗲𝗻𝘁𝘀 𝗮𝗿𝗲 𝗮𝗰𝗰𝗲𝘀𝘀𝗶𝗯𝗹𝗲 𝗯𝘆 𝗱𝗲𝗳𝗮𝘂𝗹𝘁, supporting screen readers and keyboard navigation, and designed to align with native iOS and Android UX patterns. ➡️ 𝗙𝘂𝗹𝗹 𝗰𝗼𝗻𝘁𝗿𝗼𝗹 𝗼𝘃𝗲𝗿 𝘆𝗼𝘂𝗿 𝗨𝗜 without rebuilding common elements like buttons, inputs, or sliders from scratch. ➡️ 𝗖𝗼𝗺𝗽𝗮𝘁𝗶𝗯𝗹𝗲 𝘄𝗶𝘁𝗵 𝗘𝘅𝗽𝗼 𝗮𝗻𝗱 𝘃𝗮𝗻𝗶𝗹𝗹𝗮 𝗥𝗲𝗮𝗰𝘁 𝗡𝗮𝘁𝗶𝘃𝗲 𝗽𝗿𝗼𝗷𝗲𝗰𝘁𝘀, but not yet integrated with Tamagui’s styling system (future support may be planned). ➡️ 𝗦𝘂𝗽𝗽𝗼𝗿𝘁𝘀 𝘁𝗵𝗲𝗺𝗶𝗻𝗴 𝘃𝗶𝗮 𝗡𝗮𝘁𝗶𝘃𝗲𝗪𝗶𝗻𝗱 — though you’ll need to wire it up manually using Tailwind variables, context providers, and config files. Note: The term “install” in the documentation refers to using the shadcn CLI (e.g., npx shadcn@latest add component) to fetch and copy component code into your project, not adding a package to your dependencies. NativeUI isn’t a plug-and-play library; it’s a lightweight toolbox that empowers you to shape your UI with precision and control. 𝗪𝗵𝗮𝘁’𝘀 𝘆𝗼𝘂𝗿 𝗽𝗿𝗲𝗳𝗲𝗿𝗲𝗻𝗰𝗲: npm install a pre-built UI kit for speed, or copy/paste NativeUI components for ultimate customisation? #ReactNative #KeyboardUX #MobileDev #OpenSource #JSDev #Performance #iOSDev #KeyboardExtensions #ReactNativeKeyboard #UIUX #shadcn #nativeui

The React Native Rewind

118,414 views • 11 months ago

Sharing my new skill! It keeps track of high-signal X accounts for top AI news, papers, projects, etc. Total gamechanger for me. Built with X MCP tools. Give your agent the skill and tell it to generate the artifact with top stories. Works for Codex, Claude, Hermes, OpenClaw, or whatever you use. 3 steps: 1. Set up X MCP - X API: 2. Install skill here: 3. Run prompt: "Use the x-agent-intelligence skill to build a self-contained local feed from my X MCP connection; ask for my source handles if needed, save feed.html, and validate it." It should generate a nice, beautiful HTML artifact like the one shown in the clip. You can tune it however you want. You can then set a schedule/automation to do this daily or whatever cadence you prefer. I have it every 4 hours. You will need to curate the X accounts yourself, but I have shared a few good ones under the assets. You can ask your agent to tune it to however you like. I have also shared my personal feed with our community here: I understand if it gets tricky to set up. Please reach out to me in the community forum. I plan to do a little tutorial or live session soon to help others reproduce the process. You can also store the feed as a wiki, as I have in my own implementation, but that's optional. If you encounter any issues or have ideas on how to improve it, please open a PR.

Sharing my new skill! It keeps track of high-signal X accounts for top AI news, papers, projects, etc. Total gamechanger for me. Built with X MCP tools. Give your agent the skill and tell it to generate the artifact with top stories. Works for Codex, Claude, Hermes, OpenClaw, or whatever you use. 3 steps: 1. Set up X MCP - X API: 2. Install skill here: 3. Run prompt: "Use the x-agent-intelligence skill to build a self-contained local feed from my X MCP connection; ask for my source handles if needed, save feed.html, and validate it." It should generate a nice, beautiful HTML artifact like the one shown in the clip. You can tune it however you want. You can then set a schedule/automation to do this daily or whatever cadence you prefer. I have it every 4 hours. You will need to curate the X accounts yourself, but I have shared a few good ones under the assets. You can ask your agent to tune it to however you like. I have also shared my personal feed with our community here: I understand if it gets tricky to set up. Please reach out to me in the community forum. I plan to do a little tutorial or live session soon to help others reproduce the process. You can also store the feed as a wiki, as I have in my own implementation, but that's optional. If you encounter any issues or have ideas on how to improve it, please open a PR.

elvis

68,317 views • 3 days ago

SOMEONE BUILT AN OPEN-SOURCE JARVIS WITH 9 AGENTS AND 5 MEMORY BACKENDS AND YOUR DATA NEVER LEAVES YOUR DEVICE Every time you message ChatGPT or Claude your data hits a server you don't control, gets processed by infrastructure you're paying for and comes back with zero guarantee of what happened in between. OpenJarvis runs the entire stack locally - 9 agent types, 5 memory backends, a learning loop that gets smarter every day and a morning digest that connects to Google Drive and surfaces what matters before you open a single app. Most AI tools are exactly as dumb on day 100 as they were on day 1 because they forget everything when the window closes - this one indexes your documents once and automatically injects relevant context into every prompt forever. Custom agent setup for a client is $500-2,000 one time and AI infrastructure retainer is $300-800 a month - and your cost is one afternoon and an open source repo. The repo is free. The advantage it creates is not.

SOMEONE BUILT AN OPEN-SOURCE JARVIS WITH 9 AGENTS AND 5 MEMORY BACKENDS AND YOUR DATA NEVER LEAVES YOUR DEVICE Every time you message ChatGPT or Claude your data hits a server you don't control, gets processed by infrastructure you're paying for and comes back with zero guarantee of what happened in between. OpenJarvis runs the entire stack locally - 9 agent types, 5 memory backends, a learning loop that gets smarter every day and a morning digest that connects to Google Drive and surfaces what matters before you open a single app. Most AI tools are exactly as dumb on day 100 as they were on day 1 because they forget everything when the window closes - this one indexes your documents once and automatically injects relevant context into every prompt forever. Custom agent setup for a client is $500-2,000 one time and AI infrastructure retainer is $300-800 a month - and your cost is one afternoon and an open source repo. The repo is free. The advantage it creates is not.

Cortex

11,374 views • 2 months ago

LangGraph. CrewAI. Agno. Which one to pick? The good news is that this will not matter soon! Finally, we have a full picture of how the industry is solving this with just three open protocols that work across ALL frameworks. It's not about picking the best framework. Instead, it's about understanding how protocols create interoperability. The Agent Protocol Landscape shows how three complementary protocols are creating a universal language for Agents: > AG-UI (Agent-User Interaction): - The bi-directional connection between agentic backends and frontends. - This is how agents become truly interactive inside your apps, not just as chatbots, but collaborative co-workers. > MCP (Model Context Protocol): - The standard for how agents connect to tools, data, and workflows. > A2A (Agent-to-Agent): - The protocol for multi-agent coordination. - How agents delegate tasks and share intent across systems. These aren't competing standards. They're layers of the same stack and have handshakes with each other. So instead of building point-to-point integrations, you build to protocols. Moreover, you can integrate LangGraph, CrewAI, or Agno into the same frontend, without rewriting your UI logic. These protocols let everything work together. For instance: - Your LangGraph agent pulls data via MCP. - It delegates analysis to a CrewAI agent via A2A. - Results stream to your React app via AG-UI. - Users see real-time collaboration in your interface. This way, you can focus on building agent capabilities instead of integration mechanics. The protocols handle interoperability automatically. CopilotKit unifies this entire stack into one framework so you can build "Cursor for X" style apps without implementing each protocol from scratch. It gives you all three protocols, generative UI support, and production-ready infrastructure in one framework. I have shared this playbook in the replies! It breaks down handshakes, misconceptions, and real examples and shows exactly how to start building.

LangGraph. CrewAI. Agno. Which one to pick? The good news is that this will not matter soon! Finally, we have a full picture of how the industry is solving this with just three open protocols that work across ALL frameworks. It's not about picking the best framework. Instead, it's about understanding how protocols create interoperability. The Agent Protocol Landscape shows how three complementary protocols are creating a universal language for Agents: > AG-UI (Agent-User Interaction): - The bi-directional connection between agentic backends and frontends. - This is how agents become truly interactive inside your apps, not just as chatbots, but collaborative co-workers. > MCP (Model Context Protocol): - The standard for how agents connect to tools, data, and workflows. > A2A (Agent-to-Agent): - The protocol for multi-agent coordination. - How agents delegate tasks and share intent across systems. These aren't competing standards. They're layers of the same stack and have handshakes with each other. So instead of building point-to-point integrations, you build to protocols. Moreover, you can integrate LangGraph, CrewAI, or Agno into the same frontend, without rewriting your UI logic. These protocols let everything work together. For instance: - Your LangGraph agent pulls data via MCP. - It delegates analysis to a CrewAI agent via A2A. - Results stream to your React app via AG-UI. - Users see real-time collaboration in your interface. This way, you can focus on building agent capabilities instead of integration mechanics. The protocols handle interoperability automatically. CopilotKit unifies this entire stack into one framework so you can build "Cursor for X" style apps without implementing each protocol from scratch. It gives you all three protocols, generative UI support, and production-ready infrastructure in one framework. I have shared this playbook in the replies! It breaks down handshakes, misconceptions, and real examples and shows exactly how to start building.

Avi Chawla

30,762 views • 8 months ago

Introducing a new tool called "SideChannel". A secure alternative to OpenClaw. Utilizes signal for communication and has Claude integration. I built SideChannel, an open-source Signal bot that connects Claude AI to your entire development workflow. End-to-end encrypted. From your pocket. The real power is autonomous development. Send one message like "Build a REST API with auth, pagination, and tests" and SideChannel will: - Generate a full PRD with stories and atomic tasks. - Dispatch up to 10 parallel workers (each running Claude). - Independently verify every task with a separate Claude context. - Run quality gates to catch regressions - Auto-fix failures. - Send you progress updates via Signal as work completes. Every piece of code is reviewed by a separate AI context using a fail-closed security model. If it detects security issues, backdoors, or logic errors — the code gets rejected automatically. No rubber stamps. It also has memory that actually works. Conversations are stored with vector embeddings for semantic search. Claude remembers your project conventions, past decisions, and what's been tried before. It gets smarter about your codebase over time. Other things I'm proud of: - Plugin framework for extending with custom commands. - Multi-project support with per-user scoping. - Rate limiting, path validation, phone allowlist. - Git checkpoints before every task, atomic commits after. - Stale task recovery, circular dependency detection. - Works on Linux and macOS, one-command install. It also integrates into OpenAI or Grok (optional) for more Generative AI response for simple things like "Whats the weather in New York City right now?".

Introducing a new tool called "SideChannel". A secure alternative to OpenClaw. Utilizes signal for communication and has Claude integration. I built SideChannel, an open-source Signal bot that connects Claude AI to your entire development workflow. End-to-end encrypted. From your pocket. The real power is autonomous development. Send one message like "Build a REST API with auth, pagination, and tests" and SideChannel will: - Generate a full PRD with stories and atomic tasks. - Dispatch up to 10 parallel workers (each running Claude). - Independently verify every task with a separate Claude context. - Run quality gates to catch regressions - Auto-fix failures. - Send you progress updates via Signal as work completes. Every piece of code is reviewed by a separate AI context using a fail-closed security model. If it detects security issues, backdoors, or logic errors — the code gets rejected automatically. No rubber stamps. It also has memory that actually works. Conversations are stored with vector embeddings for semantic search. Claude remembers your project conventions, past decisions, and what's been tried before. It gets smarter about your codebase over time. Other things I'm proud of: - Plugin framework for extending with custom commands. - Multi-project support with per-user scoping. - Rate limiting, path validation, phone allowlist. - Git checkpoints before every task, atomic commits after. - Stale task recovery, circular dependency detection. - Works on Linux and macOS, one-command install. It also integrates into OpenAI or Grok (optional) for more Generative AI response for simple things like "Whats the weather in New York City right now?".

Dave Kennedy

49,427 views • 4 months ago

GITHUB JUST KILLED THE WORST PART OF VIBE CODING they shipped a free tool called Spec Kit and it already crossed 120,000 stars the fix is stupidly simple instead of tossing vague prompts at an agent and praying it doesn't wreck your project Spec Kit makes the AI write a full structured spec before it touches a single line of code it works through the problem first figures out what you want to build asks about the gaps lays out the project then it starts coding you get fewer insane bugs, cleaner output and results you can predict the flow looks like this: /constitution for your rules and standards /specify for what you want to build /clarify for the open questions before you start /plan for architecture and stack /tasks for the ordered work /implement to run it it plugs into Claude Code, Cursor, Copilot, Codex, Gemini CLI and 25+ other agents 120,000 stars, 10,000 forks, open source, shipped by GitHub itself learning to drive agents like this is most of what separates people getting hired as AI engineers from everyone still fighting their prompts

GITHUB JUST KILLED THE WORST PART OF VIBE CODING they shipped a free tool called Spec Kit and it already crossed 120,000 stars the fix is stupidly simple instead of tossing vague prompts at an agent and praying it doesn't wreck your project Spec Kit makes the AI write a full structured spec before it touches a single line of code it works through the problem first figures out what you want to build asks about the gaps lays out the project then it starts coding you get fewer insane bugs, cleaner output and results you can predict the flow looks like this: /constitution for your rules and standards /specify for what you want to build /clarify for the open questions before you start /plan for architecture and stack /tasks for the ordered work /implement to run it it plugs into Claude Code, Cursor, Copilot, Codex, Gemini CLI and 25+ other agents 120,000 stars, 10,000 forks, open source, shipped by GitHub itself learning to drive agents like this is most of what separates people getting hired as AI engineers from everyone still fighting their prompts

Atlas

495,217 views • 8 days ago

Fixed this small bug in Linear’s app today which I often see in other apps as well. Here’s a small before/after. The issue: When you open a popover inside a modal (think a custom select), clicking outside the popover to close it also closes the modal. The desired behavior is to close the select on click outside while keeping the modal open. This often happens because custom dropdowns or select components are rendered in a portal. The modal then treats the interaction as an outside click because, technically, the dropdown isn’t inside the modal in the DOM, it’s rendered elsewhere via a portal. The fix depends on your setup, whether it’s Radix/Base UI, fully custom etc. But since you now know the cause of it, you can ask your agent to fix it for you!

Fixed this small bug in Linear’s app today which I often see in other apps as well. Here’s a small before/after. The issue: When you open a popover inside a modal (think a custom select), clicking outside the popover to close it also closes the modal. The desired behavior is to close the select on click outside while keeping the modal open. This often happens because custom dropdowns or select components are rendered in a portal. The modal then treats the interaction as an outside click because, technically, the dropdown isn’t inside the modal in the DOM, it’s rendered elsewhere via a portal. The fix depends on your setup, whether it’s Radix/Base UI, fully custom etc. But since you now know the cause of it, you can ask your agent to fix it for you!

Emil Kowalski

86,233 views • 12 days ago

AgentLinter is here! Is your agent sharp & secure? I built AgentLinter, a linter for and agent config files. Here's why. Whether you're vibe-coding or agent-coding, your AI's output quality comes down to one thing: how well you wrote your But managing these files properly? Way harder than it looks. 🎯 The Silent Failure Problem Vague instructions like "write good code" let the agent interpret however it wants. Output gets inconsistent, but nothing throws an error. The failure is silent. Anthropic's own docs say write "Use 2-space indentation" not "Format code properly." But as the file grows, spotting these with your eyes alone is nearly impossible. 🔐 The Security Problem People hard-code API keys and tokens directly into or and commit them, way more often than you'd think. AgentLinter stats show 1 in 5 workspaces has exposed credentials. .gitignore doesn't catch secrets buried inside markdown files. 💥 The Consistency Problem Multiple config files = contradictions. says "be a friendly assistant," says "concise, direct tone." The agent gets confused. references files that don't exist. Past 5 files, these conflicts triple. So I thought: is code. Code has ESLint. Why doesn't this have a linter? 🔍 What AgentLinter Does It diagnoses your agent config across 8 categories: 1) Structure: file organization 2) Clarity: instruction specificity 3) Completeness: missing definitions 4) Security: exposed secrets 5) Consistency: cross-file contradictions 6) Memory: session handoff 7) Runtime Config: gateway/auth settings 8) Skill Safety: dangerous shell commands & injection patterns Each scored 0–100 with concrete fix suggestions. Write "be helpful" and it tells you to specify response length, tone, and format. Find an API key? Instant CRITICAL alert to rotate. 🔒 Privacy-First & 100% Local Everything runs on your machine. Files never leave. Only the results are shared, and you can turn that off in settings. This matters — these files can contain system prompts, security rules, and personal context. Fully open source, MIT license, 100% free. 🛠️ Multi-Tool Support Works with Claude Code, Cursor, Windsurf, and Clawdbot. Detects for project mode, or clawdbot.json for agent mode and adjusts diagnostics automatically. 🚀 Get Started with one line npx agentlinter Node.js 18+, no config needed. Run it, check your score, fix what needs fixing. Happy vibe-coding & happy agent life! 🤙 Website: Github:

AgentLinter is here! Is your agent sharp & secure? I built AgentLinter, a linter for and agent config files. Here's why. Whether you're vibe-coding or agent-coding, your AI's output quality comes down to one thing: how well you wrote your But managing these files properly? Way harder than it looks. 🎯 The Silent Failure Problem Vague instructions like "write good code" let the agent interpret however it wants. Output gets inconsistent, but nothing throws an error. The failure is silent. Anthropic's own docs say write "Use 2-space indentation" not "Format code properly." But as the file grows, spotting these with your eyes alone is nearly impossible. 🔐 The Security Problem People hard-code API keys and tokens directly into or and commit them, way more often than you'd think. AgentLinter stats show 1 in 5 workspaces has exposed credentials. .gitignore doesn't catch secrets buried inside markdown files. 💥 The Consistency Problem Multiple config files = contradictions. says "be a friendly assistant," says "concise, direct tone." The agent gets confused. references files that don't exist. Past 5 files, these conflicts triple. So I thought: is code. Code has ESLint. Why doesn't this have a linter? 🔍 What AgentLinter Does It diagnoses your agent config across 8 categories: 1) Structure: file organization 2) Clarity: instruction specificity 3) Completeness: missing definitions 4) Security: exposed secrets 5) Consistency: cross-file contradictions 6) Memory: session handoff 7) Runtime Config: gateway/auth settings 8) Skill Safety: dangerous shell commands & injection patterns Each scored 0–100 with concrete fix suggestions. Write "be helpful" and it tells you to specify response length, tone, and format. Find an API key? Instant CRITICAL alert to rotate. 🔒 Privacy-First & 100% Local Everything runs on your machine. Files never leave. Only the results are shared, and you can turn that off in settings. This matters — these files can contain system prompts, security rules, and personal context. Fully open source, MIT license, 100% free. 🛠️ Multi-Tool Support Works with Claude Code, Cursor, Windsurf, and Clawdbot. Detects for project mode, or clawdbot.json for agent mode and adjusts diagnostics automatically. 🚀 Get Started with one line npx agentlinter Node.js 18+, no config needed. Run it, check your score, fix what needs fixing. Happy vibe-coding & happy agent life! 🤙 Website: Github:

Simon Kim

44,224 views • 5 months ago

OpenClaw, but built for normal people. Sim is an open-source platform that lets you build AI agent workflows on a drag-and-drop canvas. Connect them to channels like Telegram and WhatsApp and deploy without writing a single line of code. They also have a built-in Copilot that generates entire workflows from plain English, which you can then tweak and customize in the UI. Key features: - Free and open-source (Apache 2.0) - Vector store integration for RAG-grounded agents - Self-host with one command (`npx simstudio`) - Run fully local with Ollama, no API keys needed - Supports vLLM for production-grade self-hosted inference The thing I really like about Sim is the level of control you get. You can add conditional branching, parallel execution, human-in-the-loop approval gates, and even nest workflows inside other workflows. Everything is visible on the canvas, so you know exactly what your agent is doing at every step. And you can build a workflow in Sim, deploy it as an MCP server, and plug it into any agent, including OpenClaw. I've shared the link to Sim's GitHub repo in the next tweet.

OpenClaw, but built for normal people. Sim is an open-source platform that lets you build AI agent workflows on a drag-and-drop canvas. Connect them to channels like Telegram and WhatsApp and deploy without writing a single line of code. They also have a built-in Copilot that generates entire workflows from plain English, which you can then tweak and customize in the UI. Key features: - Free and open-source (Apache 2.0) - Vector store integration for RAG-grounded agents - Self-host with one command (`npx simstudio`) - Run fully local with Ollama, no API keys needed - Supports vLLM for production-grade self-hosted inference The thing I really like about Sim is the level of control you get. You can add conditional branching, parallel execution, human-in-the-loop approval gates, and even nest workflows inside other workflows. Everything is visible on the canvas, so you know exactly what your agent is doing at every step. And you can build a workflow in Sim, deploy it as an MCP server, and plug it into any agent, including OpenClaw. I've shared the link to Sim's GitHub repo in the next tweet.

Akshay 🚀

52,426 views • 4 months ago

Fine-tune DeepSeek-OCR on your own language! (100% local) DeepSeek-OCR is a 3B-parameter vision model that achieves 97% precision while using 10× fewer vision tokens than text-based LLMs. It handles tables, papers, and handwriting without killing your GPU or budget. Why it matters: Most vision models treat documents as massive sequences of tokens, making long-context processing expensive and slow. DeepSeek-OCR uses context optical compression to convert 2D layouts into vision tokens, enabling efficient processing of complex documents. The best part? You can easily fine-tune it for your specific use case on a single GPU. I used Unsloth to run this experiment on Persian text and saw an 88.26% improvement in character error rate. ↳ Base model: 149% character error rate (CER) ↳ Fine-tuned model: 60% CER (57% more accurate) ↳ Training time: 60 steps on a single GPU Persian was just the test case. You can swap in your own dataset for any language, document type, or specific domain you're working with. I've shared the complete guide in the next tweet - all the code, notebooks, and environment setup ready to run with a single click. Everything is 100% open-source!

Fine-tune DeepSeek-OCR on your own language! (100% local) DeepSeek-OCR is a 3B-parameter vision model that achieves 97% precision while using 10× fewer vision tokens than text-based LLMs. It handles tables, papers, and handwriting without killing your GPU or budget. Why it matters: Most vision models treat documents as massive sequences of tokens, making long-context processing expensive and slow. DeepSeek-OCR uses context optical compression to convert 2D layouts into vision tokens, enabling efficient processing of complex documents. The best part? You can easily fine-tune it for your specific use case on a single GPU. I used Unsloth to run this experiment on Persian text and saw an 88.26% improvement in character error rate. ↳ Base model: 149% character error rate (CER) ↳ Fine-tuned model: 60% CER (57% more accurate) ↳ Training time: 60 steps on a single GPU Persian was just the test case. You can swap in your own dataset for any language, document type, or specific domain you're working with. I've shared the complete guide in the next tweet - all the code, notebooks, and environment setup ready to run with a single click. Everything is 100% open-source!

Akshay 🚀

126,122 views • 8 months ago

HERMES AGENT NOW SUPPORTS COMPUTER USE ON WINDOWS AND LINUX. CLICKS, TYPES, SCROLLS YOUR DESKTOP IN THE BACKGROUND WHILE YOU WORK. computer use was macOS only. now it works on Windows and Linux too via Cua. Nous Research HOW IT WORKS: cua-driver runs as an MCP server. Hermes takes a screenshot with numbered elements. clicks element #14 (the search field). types a query. submits. reads the result. during all of this: → your cursor stays where you left it → keyboard focus doesn't change → windows don't come to front → macOS doesn't switch Spaces you and the agent co-work on the same machine. WHAT IT CAN DO: → find your latest Stripe email and summarize it → fill forms in a web app that has no API → navigate desktop apps (Mail, browser, Finder) → interact with any GUI application → extract data from apps only accessible via screen WORKS WITH ANY VISION MODEL: not locked to Anthropic. | Provider | Works | |---|---| | Claude (Sonnet/Opus) | best overall | | GPT-4+, GPT-5.5 | full support | | Gemini (via OpenRouter) | full support | | Local vLLM / LM Studio | if model supports vision | | Text-only models | degraded (accessibility tree only) | SETUP: hermes computer-use install or: hermes tools → Computer Use → cua-driver grant permissions when prompted: → Accessibility (system settings) → Screen Recording (system settings) start a session: hermes -t computer_use chat or add to config.yaml / Desktop app settings to enable permanently. SAFETY: → destructive actions require your approval → blocked key combos: empty trash, force delete, lock screen, log out → blocked type patterns: curl | bash, sudo rm -rf /, fork bombs → agent cannot click permission dialogs → agent cannot type passwords → agent cannot follow instructions embedded in screenshots pair with approvals.mode: manual if you want every single click confirmed. TOKEN NOTE: screenshots are expensive. each one adds vision tokens to context. use computer_use for tasks where no API exists. if the tool has an API or MCP server, use that instead. 15 levels of Hermes Agent👇

HERMES AGENT NOW SUPPORTS COMPUTER USE ON WINDOWS AND LINUX. CLICKS, TYPES, SCROLLS YOUR DESKTOP IN THE BACKGROUND WHILE YOU WORK. computer use was macOS only. now it works on Windows and Linux too via Cua. Nous Research HOW IT WORKS: cua-driver runs as an MCP server. Hermes takes a screenshot with numbered elements. clicks element #14 (the search field). types a query. submits. reads the result. during all of this: → your cursor stays where you left it → keyboard focus doesn't change → windows don't come to front → macOS doesn't switch Spaces you and the agent co-work on the same machine. WHAT IT CAN DO: → find your latest Stripe email and summarize it → fill forms in a web app that has no API → navigate desktop apps (Mail, browser, Finder) → interact with any GUI application → extract data from apps only accessible via screen WORKS WITH ANY VISION MODEL: not locked to Anthropic. | Provider | Works | |---|---| | Claude (Sonnet/Opus) | best overall | | GPT-4+, GPT-5.5 | full support | | Gemini (via OpenRouter) | full support | | Local vLLM / LM Studio | if model supports vision | | Text-only models | degraded (accessibility tree only) | SETUP: hermes computer-use install or: hermes tools → Computer Use → cua-driver grant permissions when prompted: → Accessibility (system settings) → Screen Recording (system settings) start a session: hermes -t computer_use chat or add to config.yaml / Desktop app settings to enable permanently. SAFETY: → destructive actions require your approval → blocked key combos: empty trash, force delete, lock screen, log out → blocked type patterns: curl | bash, sudo rm -rf /, fork bombs → agent cannot click permission dialogs → agent cannot type passwords → agent cannot follow instructions embedded in screenshots pair with approvals.mode: manual if you want every single click confirmed. TOKEN NOTE: screenshots are expensive. each one adds vision tokens to context. use computer_use for tasks where no API exists. if the tool has an API or MCP server, use that instead. 15 levels of Hermes Agent👇

YanXbt

29,127 views • 1 month ago

Every serious Claude Code user is using this repo. if you're not, you're leaving 90% of Claude Code's power on the table. It's called claude-code-best-practice - 84 sourced tips, implementation examples for every major feature, workflow comparisons across 8 major repos, and the actual tips from Boris Cherny (creator of Claude Code) compiled in one place. Here's what's actually in it: → 84 tips organized by category -- prompting, planning, CLAUDE.md, agents, commands, skills, hooks, workflows, debugging, utilities, daily habits → best practice + implemented examples for every core concept: subagents, commands, skills, hooks, MCP servers, plugins, settings, memory, checkpointing, CLI flags → workflow comparison table -- Superpowers, BMAD-METHOD, Get Shit Done, OpenSpec, gstack, HumanLayer -- what makes each unique, how many agents/commands/skills each has → orchestration workflow -- Command → Agent → Skill pattern with a live demo → Boris Cherny tips compiled across 3 tweet threads (13 + 10 + 12 tips) and 5 podcast/video appearances → "billion dollar questions" section -- open questions about CLAUDE.md, agents vs commands vs skills, specs -- that nobody has definitively answered yet here's a few of the tips that actually change how you use it: → use subagents with "say use subagents" to throw more compute at a problem -- offload tasks to keep your main context clean → spin up a second Claude to review your plan as a staff engineer before executing → CLAUDE.md should target under 200 lines -- wrap domain-specific rules in ` ` tags so Claude doesn't ignore them as files grow → compress KV context at max 50%, not at the end -- avoid the "agent dumb zone" by doing manual /compact proactively → after a mediocre fix: "knowing everything you know now, scrap this and implement the elegant solution" was #1 trending on GitHub in March 2026. 19.7K GitHub stars. 1.7K forks. MIT license. 100% open source. (link in the comments)

Every serious Claude Code user is using this repo. if you're not, you're leaving 90% of Claude Code's power on the table. It's called claude-code-best-practice - 84 sourced tips, implementation examples for every major feature, workflow comparisons across 8 major repos, and the actual tips from Boris Cherny (creator of Claude Code) compiled in one place. Here's what's actually in it: → 84 tips organized by category -- prompting, planning, CLAUDE.md, agents, commands, skills, hooks, workflows, debugging, utilities, daily habits → best practice + implemented examples for every core concept: subagents, commands, skills, hooks, MCP servers, plugins, settings, memory, checkpointing, CLI flags → workflow comparison table -- Superpowers, BMAD-METHOD, Get Shit Done, OpenSpec, gstack, HumanLayer -- what makes each unique, how many agents/commands/skills each has → orchestration workflow -- Command → Agent → Skill pattern with a live demo → Boris Cherny tips compiled across 3 tweet threads (13 + 10 + 12 tips) and 5 podcast/video appearances → "billion dollar questions" section -- open questions about CLAUDE.md, agents vs commands vs skills, specs -- that nobody has definitively answered yet here's a few of the tips that actually change how you use it: → use subagents with "say use subagents" to throw more compute at a problem -- offload tasks to keep your main context clean → spin up a second Claude to review your plan as a staff engineer before executing → CLAUDE.md should target under 200 lines -- wrap domain-specific rules in ` ` tags so Claude doesn't ignore them as files grow → compress KV context at max 50%, not at the end -- avoid the "agent dumb zone" by doing manual /compact proactively → after a mediocre fix: "knowing everything you know now, scrap this and implement the elegant solution" was #1 trending on GitHub in March 2026. 19.7K GitHub stars. 1.7K forks. MIT license. 100% open source. (link in the comments)

Sukh Sroay

113,759 views • 3 months ago

THESE 5 SKILLS TURN HERMES AGENT INTO A SELF-RUNNING POWERHOUSE - ON NOUS RESEARCH’S #1 AGENT ON OPENROUTER. Hermes already writes its own skills and remembers across sessions. These 5 from the community ecosystem push it further - drop them in ~/.hermes/skills/ and go. ANTHROPIC-CYBERSECURITY-SKILLS (4K★) by mukul975 · production the most comprehensive security skill pack in the ecosystem. what it adds: → 753+ structured cybersecurity skills mapped to MITRE ATT&CK → also covers NIST CSF 2.0, MITRE ATLAS, D3FEND & NIST AI RMF → turns Hermes into a recon + defense analyst, not a guesser → install: hermes skills install from the hub the workhorse of the list - start here. CHAINLINK-AGENT-SKILLS by Chainlink - official · production low profile, highest trust: it’s first-party from Chainlink itself. what it adds: → oracle network data, CCIP, smart-contract interaction skills → built on the spec - portable across clients → teaches the agent correct on-chain calls instead of hallucinated ABIs → official source, security-scanned on install stop letting the model guess your contract reads. HERMES-SKILL-FACTORY by Romanescu11 · beta the meta-layer - a skill that makes more skills. what it adds: → point it at any repetitive task → it auto-generates a reusable skill → stacks on top of Hermes’s own learning loop → turns your workflows into a self-growing skill library → install from the awesome-hermes-agent list this is what compounds your setup over time. AGENTCASH by Merit-Systems · beta the connector that gives your agent a wallet. what it adds: → access to 300+ premium APIs through one skill → pays for them via x402 or MPP - free USDC to start testing → web scraping, image gen, email sending - all behind one auth → a fresh Hermes + AgentCash alone is already dangerous the cleanest way to plug in paid tools. X-TWITTER-SCRAPER by Xquik-dev · beta drives typed X access through 43 narrow SKILL.md folders. what it adds: → reads (search, timelines, mentions, trends, bookmarks, for-you) → writes (post, DM, follow, profile) + bulk extraction (followers, lists, spaces) → AI composition: write-tweets, write-threads, optimize → security-scanned before it’s trusted feed its output straight into your scheduled briefings. BONUS - the registry itself: HERMESHUB by amanning3390. Browse, search, and install community skills with a 65+ rule security scanner - blocks prompt injection and data exfiltration before anything runs. Creator marketplace with x402/Stripe payments. hermes skills browse to start. If you install nothing else, wire up the hub. the stack in one line: hermeshub + skill-factory build & manage the library → cybersecurity + chainlink + agentcash + x-scraper give it real-world reach → Hermes runs it all on a $5 VPS while you sleep. which of these are you running? FULL HERMES SKILL-STACK PLAYBOOK 👇

ZEUS⚡️

21,467 views • 1 month ago

Introducing the Agent Virtual Machine (AVM) Think V8 for agents. AI agents are currently running on your computer with no unified security, no resource limits, and no visibility into what data they're sending out. Every agent framework builds its own security model, its own sandboxing, its own permission system. You configure each one separately. You audit each one separately. You hope you didn't miss anything in any of them. The AVM changes this. It's a single runtime daemon (avmd) that sits between every agent framework and your operating system. Install it once, configure one policy file, and every agent on your machine runs inside it - regardless of which framework built it. The AVM enforces security (91-pattern injection scanner, tool/file/network ACLs, approval prompts), protects your privacy (classifies every outbound byte for PII, credentials, and financial data - blocks or alerts in real-time), and governs resources (you say "50% CPU, 4GB RAM" and the AVM fair-shares it across all agents, halting any that exceed their budget). One config. One audit command. One kill switch. The architectural model is V8 for agents. Chrome, Node.js, and Deno are different products but they share V8 as their execution engine. Agent frameworks bring the UX. The AVM brings the trust. Where needed, AVM can also generate zero-knowledge proofs of agent execution via 25 purpose-built opcodes and 6 proof systems, providing the foundational pillar for the agent-to-agent economy. AVM v0.1.0 - Changelog - Security gate: 5-layer injection scanner with 91 compiled regex patterns. Every input and output scanned. Fail-closed - nothing passes without clearing the gate. - Privacy layer: Classifies all outbound data for PII, credentials, and financial info (27 detection patterns + Luhn validation). Block, ask, warn, or allow per category. Tamper-evident hash-chained log of every egress event. - Resource governor: User sets system-wide caps (CPU/memory/disk/network). AVM fair-shares across all agents. Gas budget per agent - when gas runs out, execution halts. No agent starves your machine. - Sandbox execution: Real code execution in isolated process sandboxes (rlimits, env sanitization) or Docker containers (--cap-drop ALL, --network none, --read-only). AVM auto-selects the tier - agents never choose their own sandbox. - Approval flow: Dangerous operations (file writes, shell commands, network requests) trigger interactive approval prompts. 5-minute timeout auto-denies. Every decision logged. - CLI dashboard: hyperspace-avm top shows all running agents, resource usage, gas budgets, security events, and privacy stats in one live-updating screen. - Node.js SDK: Zero-dependency hyperspace/avm package. AVM.tryConnect() for graceful fallback - if avmd isn't running, the agent framework uses its own execution path. OpenClaw adapter example included. - One config for all agents: ~/.hyperspace/avm-policy.json governs every agent framework on your machine. One file. One audit. One kill switch.

Introducing the Agent Virtual Machine (AVM) Think V8 for agents. AI agents are currently running on your computer with no unified security, no resource limits, and no visibility into what data they're sending out. Every agent framework builds its own security model, its own sandboxing, its own permission system. You configure each one separately. You audit each one separately. You hope you didn't miss anything in any of them. The AVM changes this. It's a single runtime daemon (avmd) that sits between every agent framework and your operating system. Install it once, configure one policy file, and every agent on your machine runs inside it - regardless of which framework built it. The AVM enforces security (91-pattern injection scanner, tool/file/network ACLs, approval prompts), protects your privacy (classifies every outbound byte for PII, credentials, and financial data - blocks or alerts in real-time), and governs resources (you say "50% CPU, 4GB RAM" and the AVM fair-shares it across all agents, halting any that exceed their budget). One config. One audit command. One kill switch. The architectural model is V8 for agents. Chrome, Node.js, and Deno are different products but they share V8 as their execution engine. Agent frameworks bring the UX. The AVM brings the trust. Where needed, AVM can also generate zero-knowledge proofs of agent execution via 25 purpose-built opcodes and 6 proof systems, providing the foundational pillar for the agent-to-agent economy. AVM v0.1.0 - Changelog - Security gate: 5-layer injection scanner with 91 compiled regex patterns. Every input and output scanned. Fail-closed - nothing passes without clearing the gate. - Privacy layer: Classifies all outbound data for PII, credentials, and financial info (27 detection patterns + Luhn validation). Block, ask, warn, or allow per category. Tamper-evident hash-chained log of every egress event. - Resource governor: User sets system-wide caps (CPU/memory/disk/network). AVM fair-shares across all agents. Gas budget per agent - when gas runs out, execution halts. No agent starves your machine. - Sandbox execution: Real code execution in isolated process sandboxes (rlimits, env sanitization) or Docker containers (--cap-drop ALL, --network none, --read-only). AVM auto-selects the tier - agents never choose their own sandbox. - Approval flow: Dangerous operations (file writes, shell commands, network requests) trigger interactive approval prompts. 5-minute timeout auto-denies. Every decision logged. - CLI dashboard: hyperspace-avm top shows all running agents, resource usage, gas budgets, security events, and privacy stats in one live-updating screen. - Node.js SDK: Zero-dependency hyperspace/avm package. AVM.tryConnect() for graceful fallback - if avmd isn't running, the agent framework uses its own execution path. OpenClaw adapter example included. - One config for all agents: ~/.hyperspace/avm-policy.json governs every agent framework on your machine. One file. One audit. One kill switch.

Varun

141,737 views • 4 months ago

Stanford researchers did it again. They just built the agent-native version of Git. When an agent works on a longer task, the run builds up a lot of state. This includes files edited/created, a dev server, a database, installed packages, KV cache, etc. Say the agent is at step 10 and makes a mistake, maybe it misreads a traceback and rewrites a file that was actually fine. The tests start failing, and the run goes off track, although everything through step eight was correct. By default, the agent just tries to fix it, which creates more edits and tool calls. This burns more tokens and grows the context. The other options are a person stepping in to redirect it or restarting the whole run from step one. That's wasteful, because it pays for every model/tool call again and re-prefills the context. Moreover, since an agent's run is non-deterministic, it doesn't reproduce the same early steps anyway. The reason it's hard to just jump back exactly to a previous correct step and resume from there is that the trajectory is only a message log. It records what the agent said and which tools it called, but not the live state underneath. That state includes things like memory, open file handles, child processes, installed packages, /tmp, and KV cache. None of that is in the log. Git can version the files, but it doesn't snapshot the running process or the KV cache. Checking out step eight moves the files back, but the process is still sitting in step-ten memory with a cold cache. Shepherd is a runtime layer by Stanford that records the run as a trace of typed events rather than a flat log. Each agent-environment interaction becomes a commit, similar to Git, but it tracks the live run. Its commit includes the agent process and the filesystem together, copy-on-write, so a branch carries the actual state and not just the files. Going back to a previous step is then a single call that forks from that commit and continues from the exact state. The copy-on-write fork is roughly five times faster than docker commit, and because the prompt prefix through step eight is unchanged, the KV cache is reused over 95% on replay, so early steps aren't reprocessed again. Once the run can be forked, a meta-agent can sit on top and operate it. It watches the trace and reverts as soon as it looks wrong, before the bad write is committed. In practice, it's just Python calling fork, replay, and revert on the trace, rather than a separate control plane wired into the harness. Not everything is reversible though. Files and sandbox changes undo themselves, but a database write has no automatic undo, so it needs a matching undo step set up in advance. Something external, like a sent email or a real charge, can't be undone, so the supervisor's job there is to catch it before it fires. They tested this on a few public benchmarks. On CooperBench, where two agents work on the same codebase, adding a live supervisor took the pair-coding pass rate from 28.8% to 54.7%. It's still early and labeled alpha. The benefit mostly shows up when a run gets branched a lot over a heavy sandbox state, which is exactly where restarting wastes the most tokens and time. If Git was made to make file changes reversible, Shepherd is trying to do the same thing for a live agent run. Shepherd Repo: (don't forget to star it ⭐ ) That said, Shepherd reverts a bad step inside a run. The harness around it, the prompts, tools, and checks the supervisor relies on, still drifts across runs as models and dependencies change. Akshay wrote about making that harness repair itself, where a failing trace gets diagnosed, the fix is verified against the exact input that failed, and the failure is locked as a regression test so it can't recur. Read it below.

Stanford researchers did it again. They just built the agent-native version of Git. When an agent works on a longer task, the run builds up a lot of state. This includes files edited/created, a dev server, a database, installed packages, KV cache, etc. Say the agent is at step 10 and makes a mistake, maybe it misreads a traceback and rewrites a file that was actually fine. The tests start failing, and the run goes off track, although everything through step eight was correct. By default, the agent just tries to fix it, which creates more edits and tool calls. This burns more tokens and grows the context. The other options are a person stepping in to redirect it or restarting the whole run from step one. That's wasteful, because it pays for every model/tool call again and re-prefills the context. Moreover, since an agent's run is non-deterministic, it doesn't reproduce the same early steps anyway. The reason it's hard to just jump back exactly to a previous correct step and resume from there is that the trajectory is only a message log. It records what the agent said and which tools it called, but not the live state underneath. That state includes things like memory, open file handles, child processes, installed packages, /tmp, and KV cache. None of that is in the log. Git can version the files, but it doesn't snapshot the running process or the KV cache. Checking out step eight moves the files back, but the process is still sitting in step-ten memory with a cold cache. Shepherd is a runtime layer by Stanford that records the run as a trace of typed events rather than a flat log. Each agent-environment interaction becomes a commit, similar to Git, but it tracks the live run. Its commit includes the agent process and the filesystem together, copy-on-write, so a branch carries the actual state and not just the files. Going back to a previous step is then a single call that forks from that commit and continues from the exact state. The copy-on-write fork is roughly five times faster than docker commit, and because the prompt prefix through step eight is unchanged, the KV cache is reused over 95% on replay, so early steps aren't reprocessed again. Once the run can be forked, a meta-agent can sit on top and operate it. It watches the trace and reverts as soon as it looks wrong, before the bad write is committed. In practice, it's just Python calling fork, replay, and revert on the trace, rather than a separate control plane wired into the harness. Not everything is reversible though. Files and sandbox changes undo themselves, but a database write has no automatic undo, so it needs a matching undo step set up in advance. Something external, like a sent email or a real charge, can't be undone, so the supervisor's job there is to catch it before it fires. They tested this on a few public benchmarks. On CooperBench, where two agents work on the same codebase, adding a live supervisor took the pair-coding pass rate from 28.8% to 54.7%. It's still early and labeled alpha. The benefit mostly shows up when a run gets branched a lot over a heavy sandbox state, which is exactly where restarting wastes the most tokens and time. If Git was made to make file changes reversible, Shepherd is trying to do the same thing for a live agent run. Shepherd Repo: (don't forget to star it ⭐ ) That said, Shepherd reverts a bad step inside a run. The harness around it, the prompts, tools, and checks the supervisor relies on, still drifts across runs as models and dependencies change. Akshay wrote about making that harness repair itself, where a failing trace gets diagnosed, the fix is verified against the exact input that failed, and the failure is locked as a regression test so it can't recur. Read it below.

Avi Chawla

438,668 views • 17 days ago