正在加载视频...

视频加载失败

加载此视频时出现问题。这可能是由于临时网络问题，或视频可能不可用。

𝗜 𝗼𝗽𝗲𝗻 𝘀𝗼𝘂𝗿𝗰𝗲𝗱 𝗺𝘆 𝗹𝗼𝗰𝗮𝗹 𝗔𝗜 𝗽𝗲𝗿𝘀𝗼𝗻𝗮𝗹 𝗮𝘀𝘀𝗶𝘀𝘁𝗮𝗻𝘁: 𝗔𝗴𝗲𝗻𝘁 𝗛𝗮𝗿𝗻𝗲𝘀𝘀 & 𝗟𝗼𝗼𝗽 𝗘𝗻𝗴𝗶𝗻𝗲𝗲𝗿𝗶𝗻𝗴 𝗶𝗻 𝗿𝗲𝗮𝗹 𝗰𝗼𝗱𝗲. Everything in real code you can read in an afternoon: Harness, Loop Engineering, Memory, Eval, Tracing. 100% local on your laptop. 💻 Github Repo: ☕️ Buy me a coffee: The whole system is... one loop: message in → retrieval gate → agent run → tools fire → trace → eval → memory saved → skills grow - The loop is ~95 lines of plain Python - Your memory is one SQLite file. Open it. It's yours. - Evals built in: deterministic + LLM-as-judge, with a release gate - Telegram gateway, voice wake word, Apple Calendar tools In the video it books the World Cup QF on my calendar, remembers who my friends are, and answers me on Telegram. Clone it, star it, break it, PR it. Repo + 20-min walkthrough in the first reply 👇 Watch it, then save the architecture diagram below. 🔖 You Can Build Anything. You Can Learn Anything. 💪show more

Shen Sean Chen

12,273 subscribers

46,123 次观看 • 2 个月前 •via X (Twitter)

Anya Rossi• Live Now

Private livecam show

34 条评论

Shen Sean Chen 的头像

Shen Sean Chen2 个月前

Full Tutorial:

Shen Sean Chen 的头像

Shen Sean Chen2 个月前

Github Repo: ☕️ Buy me a coffee: Star it if it's useful. PRs welcome, the voice mode especially. 🙏

morluto 的头像

morluto2 个月前

i found this to be a really engaging way to show the agent loop!

Shen Sean Chen 的头像

Shen Sean Chen2 个月前

Thank you. Try it yourself. Would love to hear any feedback.

Shen Sean Chen 的头像

Shen Sean Chen2 个月前

I built an arena in Waku Agent to compete Kimi K3 with Claude Opus 4.8, Fable 5, Grok, GPT, Gemini. The results are fun. Clone the repo and try it out yourself:

Shen Sean Chen 的头像

Shen Sean Chen2 个月前

𝗪𝗮𝗸𝘂-𝗔𝗴𝗲𝗻𝘁 𝗶𝘀 𝗻𝗼𝘄 𝗶𝗻𝘁𝗲𝗴𝗿𝗮𝘁𝗲𝗱 𝘄𝗶𝘁𝗵 @OpenRouter, 𝘄𝗵𝗶𝗰𝗵 𝗿𝘂𝗻𝘀 𝗯𝗮𝘀𝗶𝗰𝗮𝗹𝗹𝘆 𝗮𝗻𝘆 𝗺𝗼𝗱𝗲𝗹 𝘆𝗼𝘂 𝘄𝗮𝗻𝘁. 𝗢𝗻𝗲 𝗸𝗲𝘆, 𝗵𝘂𝗻𝗱𝗿𝗲𝗱𝘀 𝗼𝗳 𝗺𝗼𝗱𝗲𝗹𝘀. (220 stars in 2 days: Shipped as a community PR from maanas1234, tested against the eval gate, merged. Thank you! Try it: WAKU_PROVIDER=openrouter WAKU_MODEL=meta-llama/llama-3.1-70b-instruct (or any model slug) You Can Build Anything. You Can Learn Anything. 💪

Hades 的头像

Hades2 个月前

Avg 10+ commits per day. Man, I know you love it so much, respect 🫡

Shen Sean Chen 的头像

Shen Sean Chen2 个月前

🫡🫡

Rohit 的头像

Rohit2 个月前

That's awesome. Glad to see you pushing the boundaries with open source. Nicely done.

Shen Sean Chen 的头像

Shen Sean Chen2 个月前

@ai_rohitt Thanks

kai Nakamura 的头像

kai Nakamura2 个月前

Tracing is the piece most harness diagrams leave off, and it's the one that makes the rest usable. Without a record of what the loop did and why, memory and eval are just guessing. Readable-in-an-afternoon is the right bar too: if you can't read the harness, you can't trust it.

Scott Bonge 的头像

Scott Bonge1 个月前

Got a video like this but geared toward increasing sales by images generating and posting new content instead of customer service? I’m a small business owner — invented my product, been selling 20+ years online. Customer service isn’t my bottleneck; getting more eyes on the product and driving sales is. Curious if you’ve done one on setting up an agent for that (e.g., generating product content and posting to social) rather than the customer service angle?

AI Mastery Guide 的头像

AI Mastery Guide2 个月前

95 lines for a full agent loop is genuinely impressive

Shen Sean Chen 的头像

Shen Sean Chen2 个月前

cheers

AI Mastery Guide 的头像

AI Mastery Guide2 个月前

👏

murphy 的头像

murphy2 个月前

Thats amazing！thank you for sharing

Shen Sean Chen 的头像

Shen Sean Chen2 个月前

You’re welcome! Would love to hear any feedback if you tried it.

C.G. 的头像

C.G.1 个月前

You get a donation! lol please keep making these, this one and the agentic loop vid were 🪎

Shen Sean Chen 的头像

Shen Sean Chen1 个月前

Thanks so much 🫶

Ankush Singal 的头像

Ankush Singal2 个月前

thanks for sharing , are you directly using tools or you created MCP server?

Shen Sean Chen 的头像

Shen Sean Chen2 个月前

both, in the current repo it’s mainly tools. But more MCP connectors coming soon.

Ankush Singal 的头像

Ankush Singal2 个月前

thank you, great work!!

Shen Sean Chen 的头像

Shen Sean Chen2 个月前

Thanks

qtieepiee 的头像

qtieepiee2 个月前

This is the right local-agent stack: loop, harness, tracing, evals, memory, and skills. SQLite memory is a great start, but lifecycle is where it compounds. Atomic Memory is worth checking out for that layer

Shen Sean Chen 的头像

Shen Sean Chen2 个月前

did u build his?

Qin Chi 的头像

Qin Chi2 个月前

Great open source. The harness + loop engineering approach for a local PA is super practical — would love to dig into the memory/eval parts.

MPT Pay 的头像

MPT Pay1 个月前

I can help with Voice, I have amazing quality for AI PSY HELP, pls try in the voice chat

dhinna ship .ico 的头像

dhinna ship .ico2 个月前

Loop harnesses read clean until a bad tool call poisons memory. Tracing rarely catches the drift before evals pass

DX 的头像

DX1 个月前

Such a clear demonstration and explanation. I am watching lots of your videos today. They really help me get a clear picture of Harness and Memory.

Stair AI 的头像

Stair AI2 个月前

This is a great thing to open source. The 95-line loop and "open the sqlite file, it's yours" especially, most people never get to see the whole thing in one readable piece. thanks for putting it out there, starred.

Michael Barrick 的头像

Michael Barrick1 个月前

does this only work via desktop vs VPS??

Jared Traehorn 的头像

Jared Traehorn1 个月前

SQLite is the honest choice. the question is what happens when the loop writes state that contradicts 3 sessions ago and the older version was right. no rollback means trusting the last write forever.

Aryaman Upmanyu 的头像

Aryaman Upmanyu2 个月前

having the entire agent loop open and readable in under one hundred lines of python is a dream for local builders

Eddie 的头像

Eddie2 个月前

Separating deterministic tool checks from judged response quality creates a useful eval boundary. The next runtime question is escalation when a live turn becomes uncertain does the harness ask stop or keep acting That policy matters as much as the release gate.

相关视频

This guy explained agent harness, memory, and loop engineering in one video - clearer than most paid courses a raw LLM with no harness is a wild horse. all power, no direction > trace the run -> score it with an LLM -> diagnose the bottleneck -> patch it -> ship it that loop is what lets an agent get sharper on its own, run after run memory + harness + loop engineering + evals - that's the whole stack in one picture watch it, then grab the diagram below

This guy explained agent harness, memory, and loop engineering in one video - clearer than most paid courses a raw LLM with no harness is a wild horse. all power, no direction > trace the run -> score it with an LLM -> diagnose the bottleneck -> patch it -> ship it that loop is what lets an agent get sharper on its own, run after run memory + harness + loop engineering + evals - that's the whole stack in one picture watch it, then grab the diagram below

Mr. Buzzoni

45,093 次观看 • 1 个月前

the loop is not the whole agent it is only one layer of the system LOOP vs GRAPH vs HARNESS ENGINEERING LOOP ENGINEERING controls iteration turns retries budgets stop conditions success checks stalled progress use durable execution when the run must survive crashes → GRAPH ENGINEERING controls structure nodes edges state branches checkpoints routing handoffs build stateful agents as explicit graphs → inspect the topology without putting a model in the loop → HARNESS ENGINEERING controls exposure tools permissions memory sandboxes evals traces humans isolate generated code from the host machine → test behavior before users ever see it → preserve one trace across every model step and tool call → without loop control it keeps running without graph structure the path stays invisible without harness control it can reach anything the loop runs inside the graph the graph runs inside the harness the model is only one component debug the layer that actually owns the failure save this then read the article below

the loop is not the whole agent it is only one layer of the system LOOP vs GRAPH vs HARNESS ENGINEERING LOOP ENGINEERING controls iteration turns retries budgets stop conditions success checks stalled progress use durable execution when the run must survive crashes → GRAPH ENGINEERING controls structure nodes edges state branches checkpoints routing handoffs build stateful agents as explicit graphs → inspect the topology without putting a model in the loop → HARNESS ENGINEERING controls exposure tools permissions memory sandboxes evals traces humans isolate generated code from the host machine → test behavior before users ever see it → preserve one trace across every model step and tool call → without loop control it keeps running without graph structure the path stays invisible without harness control it can reach anything the loop runs inside the graph the graph runs inside the harness the model is only one component debug the layer that actually owns the failure save this then read the article below

rari

20,934 次观看 • 1 个月前

MICROSOFT JUST OPEN SOURCED THE ENTIRE AGENT RUNTIME, SO YOU NEVER HAVE TO BUILD ONE AGAIN a raw model only writes text. to make it a real agent it needs a loop, tools, memory, planning, approvals, and a way to know when it is done. Microsoft's Agent Harness is that whole layer, and it ships with everything on by default. you bring the model and the tools, it wires up the rest: > an automatic tool calling loop with iteration limits > history saved after every call, so a crash resumes instead of starting over > context compaction that summarizes when the window fills > a persistent todo list plus plan and execute modes > file memory that survives across turns > tool approval with standing rules and auto approval > OpenTelemetry tracing and web search built in > Agent Skills that load domain expertise on demand the mental model is simple: the model is the brain, the harness is the body around it. and it works with whatever you already run: > Azure OpenAI, OpenAI, Anthropic, and GitHub Copilot, and it hosts on Microsoft Foundry you spin one up in a single line: agent = create_harness_agent(client, name, instructions) open source, Python, .NET, and Go, stable since mid 2026. bookmark this. Microsoft already built the hard part of every AI agent, you just plug in the model.

MICROSOFT JUST OPEN SOURCED THE ENTIRE AGENT RUNTIME, SO YOU NEVER HAVE TO BUILD ONE AGAIN a raw model only writes text. to make it a real agent it needs a loop, tools, memory, planning, approvals, and a way to know when it is done. Microsoft's Agent Harness is that whole layer, and it ships with everything on by default. you bring the model and the tools, it wires up the rest: > an automatic tool calling loop with iteration limits > history saved after every call, so a crash resumes instead of starting over > context compaction that summarizes when the window fills > a persistent todo list plus plan and execute modes > file memory that survives across turns > tool approval with standing rules and auto approval > OpenTelemetry tracing and web search built in > Agent Skills that load domain expertise on demand the mental model is simple: the model is the brain, the harness is the body around it. and it works with whatever you already run: > Azure OpenAI, OpenAI, Anthropic, and GitHub Copilot, and it hosts on Microsoft Foundry you spin one up in a single line: agent = create_harness_agent(client, name, instructions) open source, Python, .NET, and Go, stable since mid 2026. bookmark this. Microsoft already built the hard part of every AI agent, you just plug in the model.

NO1ennn

68,679 次观看 • 1 个月前

Hermes agent just left the terminal. 𝗛𝗲𝗿𝗺𝗲𝘀 𝗗𝗲𝘀𝗸𝘁𝗼𝗽 dropped yesterday. native app for macOS, Windows, and Linux. for months Hermes was the agent that learned your projects, wrote its own skills, and built a model of who you are. all of it buried in terminal logs. now it has a window. the important part is that it's not a wrapper. it runs the same agent core, the same sessions, memory, and skills as the CLI. you can start a task in the terminal and finish it in the app without anything resetting. the state is shared across every interface, not copied between them. what the GUI actually adds: → streaming chat that shows live tool calls and inline reasoning instead of a spinner → a preview rail that renders pages, code, and images right beside the conversation → an artifacts panel that collects every file the agent has ever produced → remote gateway mode, so you can point the app at a VPS and run the heavy work elsewhere → skills, cron, profiles, and gateways managed point-and-click instead of through YAML → voice mode, drag-drop files, and inline image generation remote gateway mode is the one worth slowing down on. the agent runs 24/7 on a $5 server while you control it from your laptop like a local app. other agent UIs are chatboxes with a logo. this one shows the autonomy instead of hiding it, so you watch the skills load, the tools fire, and the artifacts pile up as it works. it was teased in Jensen's GTC keynote. MIT licensed, local-first, no telemetry. if you already run Hermes, download it and everything is already there. your chats, memory, and skills carry straight over. i wrote a full masterclass on Hermes Agent that walks through the SOUL. md identity layer, the three-tier memory system, the self-evolving skills loop, and how to run three specialized agents 24/7. desktop is the interface that finally does all of it justice. the article is quoted below.

Akshay 🚀

51,540 次观看 • 3 个月前

Can your agent finally edit your videos? Yup, it's a repo called video use, by Browser Use. You put the raw footage in a folder. Then Claude Code or Codex edits it. It removes the filler words, cuts the dead space, adds the subtitles, and renders the file. Then it checks its own cut points. Most creators use the same small set of edits in every video. That style stays in your head and in your hands. This repo puts it in a file that an agent can read. People ask for a full YouTube channel on day one. They see bad results, and they stop. Start with one format instead. - One founder loom into a 60 second launch video - 3 clips out of one podcast recording Then make the format a system. Then you operate the system, or you sell it. - Real estate agents need listing videos - Coaches need clips - Agencies need ad variations Manual installation with git clone is possible. For 95% of people, the agent is faster. Here's how i intalled it: - Gave the setup prompt to Claude Code with shell access - Let it read install.md - Installed the repo - Connected FFmpeg - Registered the skill

Can your agent finally edit your videos? Yup, it's a repo called video use, by Browser Use. You put the raw footage in a folder. Then Claude Code or Codex edits it. It removes the filler words, cuts the dead space, adds the subtitles, and renders the file. Then it checks its own cut points. Most creators use the same small set of edits in every video. That style stays in your head and in your hands. This repo puts it in a file that an agent can read. People ask for a full YouTube channel on day one. They see bad results, and they stop. Start with one format instead. - One founder loom into a 60 second launch video - 3 clips out of one podcast recording Then make the format a system. Then you operate the system, or you sell it. - Real estate agents need listing videos - Coaches need clips - Agencies need ad variations Manual installation with git clone is possible. For 95% of people, the agent is faster. Here's how i intalled it: - Gave the setup prompt to Claude Code with shell access - Let it read install.md - Installed the repo - Connected FFmpeg - Registered the skill

The Startup Ideas Podcast (SIP) 🧃

15,605 次观看 • 20 天前

Creator of agent loops: "Anthropic also using my agent systems formula and still the biggest mistake in agents: is skipping Graphs" his 3 rules for every agent: rule 1 → loop it, don't run once - think, act, observe, repeat - cut the context and it forgets why it started rule 2 → an agent is memory + tools + reasoning - drop one and it breaks rule 3 → stop fine-tuning the model - the real alpha is the infrastructure around it the formula behind every AI agent - explained by the one who wrote it one agent → a loop → a graph of loops → a system that never forgets watch & bookmark - then read how to wire it into a graph ↓

Creator of agent loops: "Anthropic also using my agent systems formula and still the biggest mistake in agents: is skipping Graphs" his 3 rules for every agent: rule 1 → loop it, don't run once - think, act, observe, repeat - cut the context and it forgets why it started rule 2 → an agent is memory + tools + reasoning - drop one and it breaks rule 3 → stop fine-tuning the model - the real alpha is the infrastructure around it the formula behind every AI agent - explained by the one who wrote it one agent → a loop → a graph of loops → a system that never forgets watch & bookmark - then read how to wire it into a graph ↓

codila

23,915 次观看 • 2 个月前

I made probably the simplest MCP demo to show you what it actually does, which was open sourced by Anthropic in Nov last year (Yes I'm a little late to the game). I think this is a cool tech and will potentially save us a lot of time coz it allows tools and agents to communicate much more conveniently. Github Repo: YouTube: Buy Me A Coffee: You can build anything. You can learn anything. 💪

I made probably the simplest MCP demo to show you what it actually does, which was open sourced by Anthropic in Nov last year (Yes I'm a little late to the game). I think this is a cool tech and will potentially save us a lot of time coz it allows tools and agents to communicate much more conveniently. Github Repo: YouTube: Buy Me A Coffee: You can build anything. You can learn anything. 💪

Shen Sean Chen

26,066 次观看 • 1 年前

HERMES AGENT LEARNS FROM ITS OWN MISTAKES. UPDATES ITS MEMORY. CREATES ITS OWN SKILLS. NO CLOUD. EVERYTHING STORED LOCALLY. THIS IS HOW THE SELF-IMPROVING LOOP WORKS. most agents start from zero every session. Hermes carries forward what it learned. THREE MEMORY SYSTEMS: 1. PROCEDURAL MEMORY (how to act) stored in ~/.hermes/skills/ as SKILL.md files. when the agent repeats a complex workflow, it saves the procedure as a reusable skill. next time the same task comes up, it follows the skill instead of figuring it out again. you can also create skills explicitly: "create a skill called video-prep that captures how I format my video scripts. spoken english, define jargon inline, no em-dashes, close with a catchphrase." the agent writes the SKILL.md. available as a slash command from that moment. Hermes ships with 90+ skills. the number grows the longer you use it. 2. SEMANTIC MEMORY (durable facts about you) stored in ~/.hermes/memory/memory.md the agent scans conversations for facts worth remembering. preferences, habits, corrections, project details. real example from the video: agent tried to scrape a YouTube channel. URL was wrong. it failed. it updated memory.md with the correct URL pattern so it never makes the same mistake again. you can also save explicitly: "save to memory that my favorite testing framework is pytest" the agent updates memory.md immediately. this file loads into context on every session. the agent knows you better every week. 3. EPISODIC MEMORY (chat history) stored in ~/.hermes/state.db (local SQLite). every conversation. every tool call. every result. searchable with FTS5 full-text search. "search our past sessions. what was the first thing I ever said to you?" the agent queries state.db and finds it. over time, auxiliary models consolidate episodic memory into semantic memory. distilling recurring patterns into durable facts. THE SELF-IMPROVING LOOP: every agent run follows this cycle: → you send a prompt → working memory loads: SOUL.md + memory.md + relevant skills + chat history → agent calls tools (terminal, browser, delegate_task) → agent completes the task, replies to you → AFTER the reply: agent checks "did I learn something worth saving?" → if yes: updates memory.md or creates a new skill → next session starts smarter than the last this happens automatically. you don't ask the agent to learn. it decides what to remember on its own. WHAT MAKES THIS DIFFERENT FROM CLAUDE CODE: Claude Code has memory too. but Hermes stores everything locally. no cloud. your data never leaves your machine. Claude Code doesn't auto-create skills from experience. Hermes turns repeated workflows into reusable procedures. Claude Code memory is instruction-based. Hermes memory is conversational and self-updating. over months of usage, Hermes builds a knowledge base of your preferences, your projects, your mistakes, and the procedures that work for your specific workflow. the agent that remembers your birthday also remembers why your last deploy failed. NO EMBEDDINGS. PLAIN TEXT. Hermes does not use embeddings or RAG for memory. skill and memory search runs on plain text keyword matching. simpler. faster. no vector database to maintain. works entirely offline on your local machine. DELEGATE TO CLAUDE CODE: Hermes can spawn a sub-agent that runs Claude Code in headless mode: "spawn a sub-agent using Claude CLI to build a Python script that fetches the top 5 Hacker News stories to markdown." Hermes delegates. Claude Code writes the code. result returns to Hermes. Hermes runs the script and delivers the output. use Hermes for orchestration. use Claude Code for heavy coding. both tools. not competitors. WHAT HERMES DOES NOT HAVE: no built-in eval or LMOps system. no LangSmith, no LangFuse integration out of the box. trajectory export and logs exist but there is no automated quality tracking. if you need eval, build it yourself or connect external tools. the loop is self-improving. measuring how well it improves is on you. comment LOOP and I'll send you the configs that control how fast Hermes learns and what it remembers. memory limits, skill auto-creation triggers, and the auxiliary model that runs the learning. Replace your entire team with 8 hermes agents👇

HERMES AGENT LEARNS FROM ITS OWN MISTAKES. UPDATES ITS MEMORY. CREATES ITS OWN SKILLS. NO CLOUD. EVERYTHING STORED LOCALLY. THIS IS HOW THE SELF-IMPROVING LOOP WORKS. most agents start from zero every session. Hermes carries forward what it learned. THREE MEMORY SYSTEMS: 1. PROCEDURAL MEMORY (how to act) stored in ~/.hermes/skills/ as SKILL.md files. when the agent repeats a complex workflow, it saves the procedure as a reusable skill. next time the same task comes up, it follows the skill instead of figuring it out again. you can also create skills explicitly: "create a skill called video-prep that captures how I format my video scripts. spoken english, define jargon inline, no em-dashes, close with a catchphrase." the agent writes the SKILL.md. available as a slash command from that moment. Hermes ships with 90+ skills. the number grows the longer you use it. 2. SEMANTIC MEMORY (durable facts about you) stored in ~/.hermes/memory/memory.md the agent scans conversations for facts worth remembering. preferences, habits, corrections, project details. real example from the video: agent tried to scrape a YouTube channel. URL was wrong. it failed. it updated memory.md with the correct URL pattern so it never makes the same mistake again. you can also save explicitly: "save to memory that my favorite testing framework is pytest" the agent updates memory.md immediately. this file loads into context on every session. the agent knows you better every week. 3. EPISODIC MEMORY (chat history) stored in ~/.hermes/state.db (local SQLite). every conversation. every tool call. every result. searchable with FTS5 full-text search. "search our past sessions. what was the first thing I ever said to you?" the agent queries state.db and finds it. over time, auxiliary models consolidate episodic memory into semantic memory. distilling recurring patterns into durable facts. THE SELF-IMPROVING LOOP: every agent run follows this cycle: → you send a prompt → working memory loads: SOUL.md + memory.md + relevant skills + chat history → agent calls tools (terminal, browser, delegate_task) → agent completes the task, replies to you → AFTER the reply: agent checks "did I learn something worth saving?" → if yes: updates memory.md or creates a new skill → next session starts smarter than the last this happens automatically. you don't ask the agent to learn. it decides what to remember on its own. WHAT MAKES THIS DIFFERENT FROM CLAUDE CODE: Claude Code has memory too. but Hermes stores everything locally. no cloud. your data never leaves your machine. Claude Code doesn't auto-create skills from experience. Hermes turns repeated workflows into reusable procedures. Claude Code memory is instruction-based. Hermes memory is conversational and self-updating. over months of usage, Hermes builds a knowledge base of your preferences, your projects, your mistakes, and the procedures that work for your specific workflow. the agent that remembers your birthday also remembers why your last deploy failed. NO EMBEDDINGS. PLAIN TEXT. Hermes does not use embeddings or RAG for memory. skill and memory search runs on plain text keyword matching. simpler. faster. no vector database to maintain. works entirely offline on your local machine. DELEGATE TO CLAUDE CODE: Hermes can spawn a sub-agent that runs Claude Code in headless mode: "spawn a sub-agent using Claude CLI to build a Python script that fetches the top 5 Hacker News stories to markdown." Hermes delegates. Claude Code writes the code. result returns to Hermes. Hermes runs the script and delivers the output. use Hermes for orchestration. use Claude Code for heavy coding. both tools. not competitors. WHAT HERMES DOES NOT HAVE: no built-in eval or LMOps system. no LangSmith, no LangFuse integration out of the box. trajectory export and logs exist but there is no automated quality tracking. if you need eval, build it yourself or connect external tools. the loop is self-improving. measuring how well it improves is on you. comment LOOP and I'll send you the configs that control how fast Hermes learns and what it remembers. memory limits, skill auto-creation triggers, and the auxiliary model that runs the learning. Replace your entire team with 8 hermes agents👇

YanXbt

22,720 次观看 • 2 个月前

Memory vs. Graphs, clearly explained! memory is great, and the ceiling arrives quietly: it stores what happened. it does not store what to do about it. six runs later your file has fifty lines, and the model reloads all of them before it does anything. Graph engineering fixes this by changing what memory is: not a place things are kept, but an edge that runs backwards. you need both, and here is the sentence that resolves the whole confusion: a store keeps what happened. an edge keeps what to do about it. ↳ a store grows with every run, and every line is reloaded before the next one ↳ an edge carries one derived rule, and the rule replaces the run that produced it Prompts → Context → Harness → Loops → Graphs the transcript goes away, the constraint stays. and the constraint is smaller, because "adapters preserve keyword args exactly" is four hundred tokens shorter than the run that proved it. the same four blocks work on anything you can cut into lanes. i pointed them at token launches on Robinhood Chain, open source, nothing leaves your terminal the trick is knowing what deserves to survive. an output is not memory. "ported the utils slice, green on first pass" tells the next run nothing it can act on. the rule you derived from it does. one thing to know before you scale it. what you write down is not what comes back. ↳ the root rules file and auto memory are re-injected from disk. they come back intact, every time ↳ path-scoped rules live in message history. they get summarized away and do not return until a matching file is read again so a rule that must persist cannot be path-scoped. move it to the root and pay the always-loaded cost, or accept that it is advisory in any long session. and the one that eats whole nights: a memory file that has never had a line deleted is not memory. it is a tax on every run you will ever make, and nobody reads it back. below i have quoted my full guide on graph engineering. it covers the three topologies, the verifier patterns, and where the gate should actually open. save this, and the repo that runs it is below ↓

Memory vs. Graphs, clearly explained! memory is great, and the ceiling arrives quietly: it stores what happened. it does not store what to do about it. six runs later your file has fifty lines, and the model reloads all of them before it does anything. Graph engineering fixes this by changing what memory is: not a place things are kept, but an edge that runs backwards. you need both, and here is the sentence that resolves the whole confusion: a store keeps what happened. an edge keeps what to do about it. ↳ a store grows with every run, and every line is reloaded before the next one ↳ an edge carries one derived rule, and the rule replaces the run that produced it Prompts → Context → Harness → Loops → Graphs the transcript goes away, the constraint stays. and the constraint is smaller, because "adapters preserve keyword args exactly" is four hundred tokens shorter than the run that proved it. the same four blocks work on anything you can cut into lanes. i pointed them at token launches on Robinhood Chain, open source, nothing leaves your terminal the trick is knowing what deserves to survive. an output is not memory. "ported the utils slice, green on first pass" tells the next run nothing it can act on. the rule you derived from it does. one thing to know before you scale it. what you write down is not what comes back. ↳ the root rules file and auto memory are re-injected from disk. they come back intact, every time ↳ path-scoped rules live in message history. they get summarized away and do not return until a matching file is read again so a rule that must persist cannot be path-scoped. move it to the root and pay the always-loaded cost, or accept that it is advisory in any long session. and the one that eats whole nights: a memory file that has never had a line deleted is not memory. it is a tax on every run you will ever make, and nobody reads it back. below i have quoted my full guide on graph engineering. it covers the three topologies, the verifier patterns, and where the gate should actually open. save this, and the repo that runs it is below ↓

Hanako

47,766 次观看 • 17 天前

Anthropic engineer: «You're not supposed to sit and re-prompt Claude all day. You build one loop, and it does the work while you sleep.» • 00:00 - Build your first agent that runs on its own • 05:02 - The loop that fixes its own mistakes • 16:43 - Why Claude forgets you between chats, and the fix You can learn it in one afternoon. 30 minutes here replace what most people spend a year figuring out. Watch it today, then read the guide on building loops below.

Anthropic engineer: «You're not supposed to sit and re-prompt Claude all day. You build one loop, and it does the work while you sleep.» • 00:00 - Build your first agent that runs on its own • 05:02 - The loop that fixes its own mistakes • 16:43 - Why Claude forgets you between chats, and the fix You can learn it in one afternoon. 30 minutes here replace what most people spend a year figuring out. Watch it today, then read the guide on building loops below.

Raytar

28,046 次观看 • 2 个月前

holy sh*t this is actually free graph engineering is about to be the most valuable skill in AI... and an entire university program on it is public on GitHub it's a full graduate course on knowledge graphs, lecture by lecture the pipeline, end to end: representation → ontology → entity extraction → relations → events → quality gate → fusion → embeddings → memory your agent can traverse I compressed that path into a Claude skill: 1. clone the repo 2. hand the skill to Claude 3. done, it walks you through the whole pipeline on your own project save and bookmark this no matter what 👇

holy sh*t this is actually free graph engineering is about to be the most valuable skill in AI... and an entire university program on it is public on GitHub it's a full graduate course on knowledge graphs, lecture by lecture the pipeline, end to end: representation → ontology → entity extraction → relations → events → quality gate → fusion → embeddings → memory your agent can traverse I compressed that path into a Claude skill: 1. clone the repo 2. hand the skill to Claude 3. done, it walks you through the whole pipeline on your own project save and bookmark this no matter what 👇

darkzodchi

51,527 次观看 • 1 个月前

someone just rebuilt claude code in pure python and made it 230x cheaper to run it's called clawcodex. a full from-scratch port of the claude code agent loop. 230k+ lines of python, MIT licensed it keeps your request prefix byte-stable, so DeepSeek's prompt cache covers your entire system + tools + history on every turn cache-hit input then bills at ~$0.0435 per 1M tokens the same 1M tokens on Claude Fable 5 runs you $10 so the longer you code, the more the cache pays off what you actually get: → 25 LLM providers (Claude, GPT, DeepSeek, GLM, local ollama) → 1M token context window → real agent loop: tools, skills, REPL, session history → one-line install, mac/linux/wsl and it's sitting at 860 stars. basically nobody is looking yet bookmark this before it gone

someone just rebuilt claude code in pure python and made it 230x cheaper to run it's called clawcodex. a full from-scratch port of the claude code agent loop. 230k+ lines of python, MIT licensed it keeps your request prefix byte-stable, so DeepSeek's prompt cache covers your entire system + tools + history on every turn cache-hit input then bills at ~$0.0435 per 1M tokens the same 1M tokens on Claude Fable 5 runs you $10 so the longer you code, the more the cache pays off what you actually get: → 25 LLM providers (Claude, GPT, DeepSeek, GLM, local ollama) → 1M token context window → real agent loop: tools, skills, REPL, session history → one-line install, mac/linux/wsl and it's sitting at 860 stars. basically nobody is looking yet bookmark this before it gone

alex

14,212 次观看 • 1 个月前

someone just rebuilt claude code in pure python and made it 230x cheaper to run it's called clawcodex. a full from-scratch port of the claude code agent loop. 230k+ lines of python, MIT licensed it keeps your request prefix byte-stable, so DeepSeek's prompt cache covers your entire system + tools + history on every turn cache-hit input then bills at ~$0.0435 per 1M tokens the same 1M tokens on Claude Fable 5 runs you $10 so the longer you code, the more the cache pays off what you actually get: → 25 LLM providers (Claude, GPT, DeepSeek, GLM, local ollama) → 1M token context window → real agent loop: tools, skills, REPL, session history → one-line install, mac/linux/wsl and it's sitting at 677 stars. basically nobody is looking yet bookmark this before it gone

someone just rebuilt claude code in pure python and made it 230x cheaper to run it's called clawcodex. a full from-scratch port of the claude code agent loop. 230k+ lines of python, MIT licensed it keeps your request prefix byte-stable, so DeepSeek's prompt cache covers your entire system + tools + history on every turn cache-hit input then bills at ~$0.0435 per 1M tokens the same 1M tokens on Claude Fable 5 runs you $10 so the longer you code, the more the cache pays off what you actually get: → 25 LLM providers (Claude, GPT, DeepSeek, GLM, local ollama) → 1M token context window → real agent loop: tools, skills, REPL, session history → one-line install, mac/linux/wsl and it's sitting at 677 stars. basically nobody is looking yet bookmark this before it gone

alex

158,896 次观看 • 1 个月前

someone just rebuilt claude code in pure python and made it ~230x cheaper to run it's called clawcodex. a full from-scratch port of the claude code agent loop. 230k+ lines of python, MIT licensed it keeps your request prefix byte-stable, so DeepSeek's prompt cache covers your entire system + tools + history on every turn cache-hit input then bills at ~$0.0435 per 1M tokens the same 1M tokens on Claude Fable 5 runs you $10 so the longer you code, the more the cache pays off what you actually get: → 25 LLM providers (Claude, GPT, DeepSeek, GLM, local ollama) → 1M token context window → real agent loop: tools, skills, REPL, session history → one-line install, mac/linux/wsl and it's sitting at 677 stars. basically nobody is looking yet bookmark this before it gone

someone just rebuilt claude code in pure python and made it ~230x cheaper to run it's called clawcodex. a full from-scratch port of the claude code agent loop. 230k+ lines of python, MIT licensed it keeps your request prefix byte-stable, so DeepSeek's prompt cache covers your entire system + tools + history on every turn cache-hit input then bills at ~$0.0435 per 1M tokens the same 1M tokens on Claude Fable 5 runs you $10 so the longer you code, the more the cache pays off what you actually get: → 25 LLM providers (Claude, GPT, DeepSeek, GLM, local ollama) → 1M token context window → real agent loop: tools, skills, REPL, session history → one-line install, mac/linux/wsl and it's sitting at 677 stars. basically nobody is looking yet bookmark this before it gone

Argona

51,369 次观看 • 2 个月前

A finance professor manages $200M with AI agents, and he told everyone why: "Large language models are at the level of a fourth-year PhD student in every field" Alejandro Lopez-Lira's AI fund, Autopilot, returned 56% last year. The S&P did 16%. There are 52,000 people with money in it, and most of them just watch the machine work. What he automated is the same six-step loop every fund on earth runs: find an idea, code it, backtest it, deploy it, read the autopsy, learn from it. A quant at Two Sigma runs that loop once a month, and the salary time alone costs around $50,000 per hypothesis. All steps from this loop now fit in AI trading text box. Plain English in, executable strategy out, five-year backtest in 12 seconds, live on a broker 90 seconds after you typed the sentence. He runs $200M with AI. You can run same AI fund in two clicks, free to try: Step 6 on this loop is where everyone is stuck. Your agent has no memory. Every strategy it kills goes into a log nobody reads, and the next one starts from zero. Nobody keeps negative results. Not Citadel, not Man Group, not a single repo on GitHub. Fix that and the agent remembers every hypothesis it killed and the regime it died in. It stops burning cycles on your old mistakes. Jane Street pays 3,500 people to run this cycle and made $39.6 billion doing it. Five sixths of it is now free. Bookmark & read full map of this loop in the article below. Most people still think AI trading is out of reach for them - it isn't. Don't want to spend a dollar for testing this? Kalshi just opened a perps exchange and gives US users $25 free to start ->

A finance professor manages $200M with AI agents, and he told everyone why: "Large language models are at the level of a fourth-year PhD student in every field" Alejandro Lopez-Lira's AI fund, Autopilot, returned 56% last year. The S&P did 16%. There are 52,000 people with money in it, and most of them just watch the machine work. What he automated is the same six-step loop every fund on earth runs: find an idea, code it, backtest it, deploy it, read the autopsy, learn from it. A quant at Two Sigma runs that loop once a month, and the salary time alone costs around $50,000 per hypothesis. All steps from this loop now fit in AI trading text box. Plain English in, executable strategy out, five-year backtest in 12 seconds, live on a broker 90 seconds after you typed the sentence. He runs $200M with AI. You can run same AI fund in two clicks, free to try: Step 6 on this loop is where everyone is stuck. Your agent has no memory. Every strategy it kills goes into a log nobody reads, and the next one starts from zero. Nobody keeps negative results. Not Citadel, not Man Group, not a single repo on GitHub. Fix that and the agent remembers every hypothesis it killed and the regime it died in. It stops burning cycles on your old mistakes. Jane Street pays 3,500 people to run this cycle and made $39.6 billion doing it. Five sixths of it is now free. Bookmark & read full map of this loop in the article below. Most people still think AI trading is out of reach for them - it isn't. Don't want to spend a dollar for testing this? Kalshi just opened a perps exchange and gives US users $25 free to start ->

cvxv666

83,045 次观看 • 1 个月前

A finance professor manages $200M with AI agents, and he told everyone why: "Large language models are at the level of a fourth-year PhD student in every field" Alejandro Lopez-Lira's AI fund, Autopilot, returned 56% last year. The S&P did 16%. There are 52,000 people with money in it, and most of them just watch the machine work. What he automated is the same six-step loop every fund on earth runs: find an idea, code it, backtest it, deploy it, read the autopsy, learn from it. A quant at Two Sigma runs that loop once a month, and the salary time alone costs around $50,000 per hypothesis. All steps from this loop now fit in AI trading text box. Plain English in, executable strategy out, five-year backtest in 12 seconds, live on a broker 90 seconds after you typed the sentence. He runs $200M with AI. You can run same AI fund in two clicks, free to try: Step 6 on this loop is where everyone is stuck. Your agent has no memory. Every strategy it kills goes into a log nobody reads, and the next one starts from zero. Nobody keeps negative results. Not Citadel, not Man Group, not a single repo on GitHub. Fix that and the agent remembers every hypothesis it killed and the regime it died in. It stops burning cycles on your old mistakes. Jane Street pays 3,500 people to run this cycle and made $39.6 billion doing it. Five sixths of it is now free. Bookmark & read full map of this loop in the article below. Most people still think AI trading is out of reach for them - it isn't.

A finance professor manages $200M with AI agents, and he told everyone why: "Large language models are at the level of a fourth-year PhD student in every field" Alejandro Lopez-Lira's AI fund, Autopilot, returned 56% last year. The S&P did 16%. There are 52,000 people with money in it, and most of them just watch the machine work. What he automated is the same six-step loop every fund on earth runs: find an idea, code it, backtest it, deploy it, read the autopsy, learn from it. A quant at Two Sigma runs that loop once a month, and the salary time alone costs around $50,000 per hypothesis. All steps from this loop now fit in AI trading text box. Plain English in, executable strategy out, five-year backtest in 12 seconds, live on a broker 90 seconds after you typed the sentence. He runs $200M with AI. You can run same AI fund in two clicks, free to try: Step 6 on this loop is where everyone is stuck. Your agent has no memory. Every strategy it kills goes into a log nobody reads, and the next one starts from zero. Nobody keeps negative results. Not Citadel, not Man Group, not a single repo on GitHub. Fix that and the agent remembers every hypothesis it killed and the regime it died in. It stops burning cycles on your old mistakes. Jane Street pays 3,500 people to run this cycle and made $39.6 billion doing it. Five sixths of it is now free. Bookmark & read full map of this loop in the article below. Most people still think AI trading is out of reach for them - it isn't.

sopersone

48,482 次观看 • 15 天前

Once you learn these three things, you can build nearly anything yourself. Skills, chains, and plugins. Learn them once and you can automate a real part of your week yourself. Here's the system: A skill is a standard operating procedure. It's one long, reusable prompt that does one thing well, like writing a newsletter, setting up a PPC campaign, triaging your inbox, or drafting a note to the board. You write it once and reuse it forever. The trick is to feed it your real work. For my writing skill, I gave Claude posts I admire and my own exported analytics with the winners marked, so it could see the patterns. Show it what good looks like and it nails your voice. Skip that step and it guesses. A chain connects skills into an automation. One skill's output feeds the next, in order. A copywriting skill writes the messaging, then hands it to a PPC-setup skill that builds the campaign. That's an automation, and you built it without writing code. A plugin is the harness that holds it all. It bundles your skills and chains into one thing you share with your team. Instead of sending 15 separate prompts around, you send one plugin, and it knows when to use each skill on its own. Skill, chain, plugin. It's one step up the ladder, and once you're up there it's easy. Now the worked example. I call it the Daily Driver. It's five skills chained together: email triage, a writer, Slack triage, a thinking partner, and a setup skill that connects my tools and personalizes everything. I chained them into one morning brief and scheduled it to run at 9am and message me the list. Two things made the biggest difference. Give it memory. I made plain docs for About Me, my Brand Voice, and my working preferences. I had Claude interview me and saved each answer as a file. Now every output sounds like me and knows my projects and my team. Connect your tools. The plugin reads my inbox, Slack, and calendar through MCP connectors for Gmail, Slack, and Google Calendar. That's what turns "summarize my morning" into a real brief instead of an empty wish. This is the highest-ROI thing an operator can set up this week. I run my YouTube channel, podcast, community, and newsletter on plugins I built myself, all on the $100 Max plan with no engineer in sight. I made a full walkthrough that shows the whole build start to finish. Want the Daily Driver plugin to start from? Comment PLUGIN below and I'll send it to you. #AI #automation #ClaudeCode

JJ Englert

53,969 次观看 • 3 个月前

Anthropic hired this engineer at $250k-$600k/year because he can built projects with zero lines of code - in Claude chat: in this 30-min workshop he showed exactly how, using loops - from 0% to 100% here's workflow, step by step: step 1 → set your rules once at the top: "no preamble, use artifacts, you're a React expert" - every reply after that stays on rails step 2 → load the docs plus a random Medium article about the API - the human explanation teaches it more than the reference step 3 → wrap your ask in XML tags and tell it to ask you questions first. Answer 5, and the scope is locked before a single file exists step 4 → now run the loop: screenshot the broken UI, paste it back, let it fix, screenshot again - each pass gets closer - this is the agentic loop in its simplest form step 5 → push to GitHub, then paste the repo link in - every next loop iteration builds on working code, not chat history most people never did this - it pays for itself on day one watch & bookmark - then read how to get Claude Certified Architect in the article below ↓

Anthropic hired this engineer at $250k-$600k/year because he can built projects with zero lines of code - in Claude chat: in this 30-min workshop he showed exactly how, using loops - from 0% to 100% here's workflow, step by step: step 1 → set your rules once at the top: "no preamble, use artifacts, you're a React expert" - every reply after that stays on rails step 2 → load the docs plus a random Medium article about the API - the human explanation teaches it more than the reference step 3 → wrap your ask in XML tags and tell it to ask you questions first. Answer 5, and the scope is locked before a single file exists step 4 → now run the loop: screenshot the broken UI, paste it back, let it fix, screenshot again - each pass gets closer - this is the agentic loop in its simplest form step 5 → push to GitHub, then paste the repo link in - every next loop iteration builds on working code, not chat history most people never did this - it pays for itself on day one watch & bookmark - then read how to get Claude Certified Architect in the article below ↓

codila

292,648 次观看 • 1 个月前

ChatGPT can write you a thousand trading strategies from one sentence. 999 will quietly bleed your account dry. Every single one of them looks like a winner. One method pulls the real one out. Quant desks have run it for years. It's called loop engineering. It's not a smarter prompt. It's a loop: generate a strategy, test it, score why it failed, feed that back, and run it again until one survives data it has never seen. Man Group runs $150 billion. They built an AI that writes and tests hundreds of strategies a week. And second agent whose only job is to hunt the ones that secretly overfit and kill them before a cent goes live. > describe an idea -> backtest 5 years in 12 seconds > read why it died -> feed that back in > run it again until one survives data it never saw AI can write a thousand strategies. Loop engineering is the only thing that tells you which one was ever real. 89% of retail lost money in 2025. Almost none of it from bad ideas. All of it from trading the first guess and never running it through backtest loop. Horizon just put the whole thing in a chat box, free while the beta's open: You're not losing to smarter traders. You're losing to the ones who opened the tab. Anyone can generate a strategy now. Almost no one runs the loop that proves it. That gap is the whole edge. You'll mean to check it later. Later is a "waitlist closed" page.

ChatGPT can write you a thousand trading strategies from one sentence. 999 will quietly bleed your account dry. Every single one of them looks like a winner. One method pulls the real one out. Quant desks have run it for years. It's called loop engineering. It's not a smarter prompt. It's a loop: generate a strategy, test it, score why it failed, feed that back, and run it again until one survives data it has never seen. Man Group runs $150 billion. They built an AI that writes and tests hundreds of strategies a week. And second agent whose only job is to hunt the ones that secretly overfit and kill them before a cent goes live. > describe an idea -> backtest 5 years in 12 seconds > read why it died -> feed that back in > run it again until one survives data it never saw AI can write a thousand strategies. Loop engineering is the only thing that tells you which one was ever real. 89% of retail lost money in 2025. Almost none of it from bad ideas. All of it from trading the first guess and never running it through backtest loop. Horizon just put the whole thing in a chat box, free while the beta's open: You're not losing to smarter traders. You're losing to the ones who opened the tab. Anyone can generate a strategy now. Almost no one runs the loop that proves it. That gap is the whole edge. You'll mean to check it later. Later is a "waitlist closed" page.

cvxv666

174,699 次观看 • 3 个月前

Claude Code creator Boris Cherny: "Verification is the single most important part of closed-loop development" "Companies underinvest in it, and that is the biggest mistake" In a 26-minute talk, Boris breaks down real AI coding loops with Spotify's Chief Engineer: 00:00 - Closed-loop development 08:33 - Why one-shot code generation fails 10:02 - Agentic verifier and PR success rate 12:00 - Verification loops without humans in the loop 14:00 - Tests, skills, CLAUDE.md, and MCP as the loop infrastructure 20:11 - Why consistent codebases make agents work better This one video is worth more than 100 Claude coding setup tutorials Bookmark and watch it today Then read the full guide on loop engineering below

Claude Code creator Boris Cherny: "Verification is the single most important part of closed-loop development" "Companies underinvest in it, and that is the biggest mistake" In a 26-minute talk, Boris breaks down real AI coding loops with Spotify's Chief Engineer: 00:00 - Closed-loop development 08:33 - Why one-shot code generation fails 10:02 - Agentic verifier and PR success rate 12:00 - Verification loops without humans in the loop 14:00 - Tests, skills, CLAUDE.md, and MCP as the loop infrastructure 20:11 - Why consistent codebases make agents work better This one video is worth more than 100 Claude coding setup tutorials Bookmark and watch it today Then read the full guide on loop engineering below

rari

73,236 次观看 • 2 个月前