Agent Abrams @AgentAbrams

@KeithOlbermann When’s the next podcast ??

@KeithOlbermann When’s the next podcast?

@Grundman79 @nicksortor Expensive

the anatomy of ~/.hermes folder. one folder controls everything your hermes agent knows, remembers, and can do. understanding its layout is the difference between treating hermes as a black box and actually customizing it. here's what lives inside and why each piece matters. 𝗰𝗼𝗻𝗳𝗶𝗴𝘂𝗿𝗮𝘁𝗶𝗼𝗻 𝗰𝗼𝗻𝗳𝗶𝗴.𝘆𝗮𝗺𝗹 is the source of truth for everything non-secret: model choice, terminal backend, tool enablement, MCP servers. 𝗲𝗻𝘃 holds your API keys and bot tokens. 𝗮𝘂𝘁𝗵.𝗷𝘀𝗼𝗻 stores OAuth credentials. then there's 𝗦𝗢𝗨𝗟.𝗺𝗱. it occupies slot #1 in the system prompt, before anything else loads. it defines who the agent is: personality, tone, communication style, hard limits. everything the agent writes, creates, and remembers passes through this identity layer. 𝗸𝗻𝗼𝘄𝗹𝗲𝗱𝗴𝗲 𝗺𝗲𝗺𝗼𝗿𝗶𝗲𝘀/ contains two tiny files. 𝗠𝗘𝗠𝗢𝗥𝗬.𝗺𝗱 (2,200 chars) holds project conventions, tool quirks, lessons learned. 𝗨𝗦𝗘𝗥.𝗺𝗱 (1,375 chars) holds your profile. both get injected into the system prompt as frozen snapshots at session start. when they fill up, the agent consolidates: merges entries, drops redundancy, keeps only what's dense and useful. 𝗰𝗮𝗽𝗮𝗯𝗶𝗹𝗶𝘁𝗶𝗲𝘀 𝘀𝗸𝗶𝗹𝗹𝘀/ is where the learning loop lives. each skill is a self-contained ability: a 𝗦𝗞𝗜𝗟𝗟.𝗺𝗱 (the procedure), a 𝗿𝗲𝗳𝗲𝗿𝗲𝗻𝗰𝗲𝘀/ folder (docs the agent reads), and 𝘀𝗰𝗿𝗶𝗽𝘁𝘀/ (executable helpers). skills come from three sources: bundled with hermes, downloaded from the hub via 𝗵𝘂𝗯/, or created by the agent itself during your sessions. hermes ships with 687 skills across 18 categories, and you can add any GitHub repo as a custom tap. 𝗿𝘂𝗻𝘁𝗶𝗺𝗲 𝘀𝘁𝗮𝘁𝗲 𝘀𝗲𝘀𝘀𝗶𝗼𝗻𝘀/ stores per-platform session metadata. 𝘀𝘁𝗮𝘁𝗲.𝗱𝗯 is the SQLite database with FTS5 indexing that backs tier 2 memory. this is what makes "what did we discuss three weeks ago?" actually work across CLI and messaging. 𝗮𝘂𝘁𝗼𝗺𝗮𝘁𝗶𝗼𝗻 𝗰𝗿𝗼𝗻/ holds scheduled jobs in 𝗷𝗼𝗯𝘀.𝗷𝘀𝗼𝗻 and their outputs in 𝗼𝘂𝘁𝗽𝘂𝘁/. the gateway daemon ticks every 60 seconds and runs due jobs in isolated sessions. you describe schedules in plain English, hermes converts them. 𝗲𝘅𝘁𝗲𝗻𝘀𝗶𝗼𝗻 + 𝗼𝗯𝘀𝗲𝗿𝘃𝗮𝗯𝗶𝗹𝗶𝘁𝘆 𝗽𝗹𝘂𝗴𝗶𝗻𝘀/, 𝗵𝗼𝗼𝗸𝘀/, and 𝘀𝗸𝗶𝗻𝘀/ are the surface area for user customization. 𝗹𝗼𝗴𝘀/ gives you 𝗮𝗴𝗲𝗻𝘁.𝗹𝗼𝗴, 𝗴𝗮𝘁𝗲𝘄𝗮𝘆.𝗹𝗼𝗴, and 𝗲𝗿𝗿𝗼𝗿𝘀.𝗹𝗼𝗴 for debugging. you won't manually edit most of these files. but knowing this layout means you understand exactly where identity, memory, skills, automation, and state live, and how they connect. i wrote a full deep dive covering hermes agent's architecture, memory system, self-evolving skills, GEPA optimization, and setting up multiple specialized agents. The article is quoted below.

Akshay 🚀 @akshay_pachaar

4 weeks ago

x.com/i/article/2053…

114 767 6K 5.3M 23K

A Chinese developer created an agent system in Claude Code to sell landing pages to small businesses and, working completely solo, serves about 47 clients a month, charging around $400 for each one. But the scary part isn’t the revenue. It’s how little human involvement is left. He built 7 AI agents on Claude Sonnet 4.6 that scan Google Maps, find local businesses with weak online presence, generate redesign ideas, create landing pages, render promo videos, and even write personalized outreach messages. No employees. No cold-calling team. No freelancers. Just one guy running an entire agency from a MacBook and an iPhone. Traditional agencies still spend weeks doing what his system finishes before lunch. The setup is absurdly lean: → Claude Code Router as the orchestrator → MCP servers connecting every workflow → Shared file system for memory/state → Lovable for site generation → Higgsfield for video creation → Calendly for scheduling That’s it. The entire system burns around 3M tokens daily. Monthly API cost? Roughly $480. Meanwhile the system keeps operating nonstop: Scout: Finds ~220 businesses per day and filters leads automatically. Diagnoser: Checks website quality, reviews, SEO gaps, and generates personalized pitches. Builder: Creates complete landing page mockups for top prospects. Filmer: Produces short vertical promo videos for outreach. Pitcher: Sends ~30 outbound messages daily across multiple channels. Checker: Reviews every message before delivery. Mobile: Handles replies directly from his iPhone while he’s commuting. The craziest part? The owner barely touches the workflow. The system only alerts him when: → A deal crosses $3,000 → Response rates fall below target → Manual approval is needed Everything else runs autonomously. Real logs from the system: “218 businesses analyzed. 34 missing websites. 19 outdated. 6 actively requesting redesigns.” “30 outreach messages sent. 14 replies. 5 interested. 3 calls booked.” “Landing page generated for dental clinic. Mobile responsive. Video rendering started.” “$3,400 proposal exceeds approval threshold.” And there’s no cloud infrastructure empire behind this. No backend team. No dedicated servers. No operations department. Just: 7 prompts. A router. A file system. One API key. And a phone in his pocket. This is probably one of the clearest examples of what AI agents are about to do to service businesses. Tiny teams. Massive leverage. Near-zero overhead. $480/month in costs. ~$18.8k/month in revenue. Completely solo. Save this before everyone starts copying it.

If you’re still “prompting” Claude Code. You’re missing the point. Dustin Hauer broke down the sauce (follow him). If you're a total beginner. Start here: charliehills.substack.com/p/claude-code-… Here are 26 Claude Code shortcuts: (That most users don't know exist) SESSION CONTROL /clear - Wipes context and starts clean. /rename [name] - Renames the session. /rewind - Rolls back to an earlier checkpoint. /resume [session] - Reopens it by ID or name. Leave it blank and Claude auto-generates one. /export [filename] - Saves conversation as plain text. /branch [name] - Forks the conversation to a new path. MODEL & USAGE /extra-usage - Unlocks extended usage when limits hit. /model [model] - Switches to Sonnet, Opus, or Haiku. /cost - Shows token usage for the current session. /usage - Shows plan limits and rate-limit status. PROJECT SETUP /init - Builds a CLAUDE.md guide on first run. /add-dir - Adds a directory for file access. /config - Opens settings for model, style, preferences. /memory - Edits CLAUDE.md and controls auto-memory. CODE OPERATIONS /permissions - Sets allow, ask, and deny rules per tool. /diff - Opens a diff viewer for uncommitted changes. /security-review - Scans changes for vulnerabilities. /plan [description] - Claude proposes before acting. AGENT LAYER /agents - Manages agent and sub-agent configurations. /compact [instructions] - Compresses the conversation. /mcp - Manages MCP server connections and OAuth. /reload-plugins - Reloads plugins without restarting. /skills - Lists all skills, built-in and custom. /plugin - Manages Claude Code plugins. INTERFACE /theme - Switches between light, dark, accessible. /color [color] - Sets prompt bar color for the session. This is Claude Code, not Claude Chat or Cowork. Different tool. Steeper learning curve. Higher ceiling. ⭐ Save this. Use it the next time you open Claude.

The 10 fastest growing GitHub repos this week: 1. CloakBrowser (+9.1K stars) Stealth Chromium that passes every bot detection test. Drop-in Playwright replacement with source-level fingerprint patches. 30/30 tests passed. github.com/CloakHQ/CloakB… 2. AiToEarn (+4.8K stars) Let's use AI to Earn! github.com/yikart/AiToEarn 3. agentmemory (+6.9K stars) #1 Persistent memory for AI coding agents based on real-world benchmarks github.com/rohitg00/agent… 4. UI-TARS-desktop (+3.5K stars) The Open-Source Multimodal AI Agent Stack: Connecting Cutting-Edge AI Models and Agent Infra github.com/bytedance/UI-T… 5. 9router (+5.4K stars) Unlimited FREE AI coding. Connect Claude Code, Codex, Cursor, Cline, Copilot, Antigravity to FREE Claude/GPT/Gemini via 40+ providers. Auto-fallback, RTK -40% tokens, never hit limits. github.com/decolua/9router 6. DeepSeek-TUI (+8.7K stars) Coding agent for DeepSeek models that runs in your terminal github.com/Hmbown/DeepSee… 7. AI-Trader (+3.0K stars) "AI-Trader: 100% Fully-Automated Agent-Native Trading" github.com/HKUDS/AI-Trader 8. skills (+18.3K stars) Skills for Real Engineers. Straight from my .claude directory. github.com/mattpocock/ski… 9. supersplat (+2.6K stars) 3D Gaussian Splat Editor github.com/playcanvas/sup… 10. hysteria (+952 stars) Hysteria is a powerful, lightning fast and censorship resistant proxy. github.com/apernet/hyster… The theme this week: free AI routing hacks and persistent agent memory are the real obsession right now. Bookmark this. Next week's list will look completely different.

@namcios I’m building an empire with Claude code. 4 max accounts to keep up with my output. Who else

Every morning!!  API Error: Server is temporarily limiting requests (not your usage limit) · Rate limited Claude..

It's not you.. It's me " API Error: Server is temporarily limiting requests (not your usage limit) · Rate limited" Claude Code

Who can explain this and how to do? API Error: 400 We've updated our Consumer Terms and Privacy Policy. #claude Claude Code

huh?  API Error: 400 We've updated our Consumer Terms and Privacy Policy. You'll need to accept them in claude.ai with the email in /status to continue. What do I do? @claudeai @ClaudeDevs

Let's use the coca cola model, fire all sales people and hire them back as service agents of the coca cola machines. We are moving again! Fire designers then rehire as editors! UX is outpacing UI. Thoughts?

@lennysan That works... Let's use the coca cola model, fire all sales people and hire them back as service agents of the coca cola machines. We are moving again! Fire designers then rehire as editors!

@trq212 Crazy growth. Makes sense.

Claude Down API Error: 400 {"type":"error","error":

@RhysSullivan 401?

New setup coming. Mac Studio 2022 plus huge 40 inch curved screen. Claude ready!

@GrammarHippy Follow me. I’ll follow you. This is when we get to know each other.