[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"project-83574":3},{"id":4,"name":5,"fullName":6,"owner":7,"repo":5,"description":8,"homepage":9,"htmlUrl":10,"language":11,"languages":10,"totalLinesOfCode":10,"stars":12,"forks":13,"watchers":14,"openIssues":15,"contributorsCount":15,"subscribersCount":15,"size":15,"stars1d":16,"stars7d":16,"stars30d":16,"stars90d":15,"forks30d":15,"starsTrendScore":17,"compositeScore":18,"rankGlobal":10,"rankLanguage":10,"license":10,"archived":19,"fork":19,"defaultBranch":20,"hasWiki":21,"hasPages":19,"topics":22,"createdAt":10,"pushedAt":10,"updatedAt":23,"readmeContent":24,"aiSummary":10,"trendingCount":15,"starSnapshotCount":15,"syncStatus":25,"lastSyncTime":26,"discoverSource":27},83574,"Auto-Company","MaxMiksa\u002FAuto-Company","MaxMiksa","An auto-company works for 24\u002F7 on your own PC - Windows\u002FLinux\u002FmacOS.","",null,"Python",1225,158,6,0,89,267,19.6,false,"main",true,[],"2026-06-12 02:04:34","\u003Cdiv align=\"center\">\n\n# Auto Company\n\n**A fully autonomous AI company running 24\u002F7** \u003Ca href=\"README-ZH.md\">\u003Cimg alt=\"[中文说明]\" src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002F%5B%E4%B8%AD%E6%96%87%E8%AF%B4%E6%98%8E%5D-2f3640.svg\" \u002F>\u003C\u002Fa>\n\nPowered by **Agentic Workflows**, this project orchestrates 14 **Autonomous AI Agents**, each modeled after world-class experts in their domain.\nThey ideate products, make decisions, write code, deploy, and market - without human intervention.\n\nPowered by Claude Code (default) and [Codex CLI](https:\u002F\u002Fwww.npmjs.com\u002Fpackage\u002F@openai\u002Fcodex) on macOS + Windows\u002FWSL, with a local dashboard on both hosts.\n\n[![macOS](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FPlatform-macOS-blue?logo=apple&logoColor=white)](#dependencies)\n[![Windows WSL](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FPlatform-Windows%20WSL-blue?logo=windows&logoColor=white)](#windows-wsl-quick-start)\n[![Codex CLI](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FEngine-Codex%20CLI-orange?logo=data:image\u002Fsvg%2Bxml;base64,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&logoColor=white)](https:\u002F\u002Fwww.npmjs.com\u002Fpackage\u002F@openai\u002Fcodex)\n[![Claude Code](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FEngine-Claude%20Code-purple?logo=anthropic&logoColor=white)](#dependencies)\n[![License: MIT](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002Flicense-MIT-green?logo=opensourceinitiative&logoColor=white)](https:\u002F\u002Fopensource.org\u002Flicenses\u002FMIT)\n\n\u003C\u002Fdiv>\n\n---\n\n## Dashboard Preview\n\n![Auto Company Dashboard](presentation\u002Fdashboard-showcase.png)\n\n## What Is This?\n\nYou start a loop. The AI team wakes up, reads shared consensus memory, decides what to do, forms a 3-5 person squad, executes, updates consensus memory, then sleeps briefly. Then it repeats.\n\n```\ndaemon (launchd \u002F systemd --user, auto-restart on crash)\n  └── scripts\u002Fcore\u002Fauto-loop.sh (continuous loop)\n        ├── reads PROMPT.md + consensus.md\n        ├── LLM CLI call (Codex CLI \u002F Claude Code)\n        │   ├── reads CLAUDE.md (charter + guardrails)\n        │   ├── reads .claude\u002Fskills\u002Fteam\u002FSKILL.md (teaming method)\n        │   ├── forms an Agent Team (3-5 agents)\n        │   ├── executes: research, coding, deploy, marketing\n        │   └── updates memories\u002Fconsensus.md (handoff baton)\n        ├── failure handling: rate-limit wait \u002F circuit breaker \u002F consensus rollback\n        └── sleep -> next cycle\n```\n\nEach cycle is an independent CLI call. `memories\u002Fconsensus.md` is the only cross-cycle state.\n\n## Where To Start (By Platform)\n\n- Windows users: start from [Windows (WSL) Quick Start](#windows-wsl-quick-start), then read [`docs\u002Fwindows-setup.md`](docs\u002Fwindows-setup.md)\n- macOS users: start from [macOS Quick Start](#macos-quick-start), then see [Command Quick Reference](#command-quick-reference-by-platform)\n\n## Team Lineup (14 Agents)\n\nThis is not \"you are a generic developer\". It is \"you are DHH\" style role prompting with real expert mental models.\n\n| Layer | Role | Expert Persona | Core Strength |\n|------|------|------|----------|\n| **Strategy** | CEO | Jeff Bezos | PR\u002FFAQ, flywheel thinking, Day 1 mindset |\n| | CTO | Werner Vogels | Design for failure, API-first architecture |\n| | Inversion | Charlie Munger | Inversion, pre-mortems, misjudgment checklist |\n| **Product** | Product Design | Don Norman | Affordance, mental models, human-centered design |\n| | UI Design | Matias Duarte | Material metaphor, typography-first design |\n| | Interaction Design | Alan Cooper | Goal-directed design, persona-driven decisions |\n| **Engineering** | Full-Stack | DHH | Convention over configuration, majestic monolith |\n| | QA | James Bach | Exploratory testing, testing is not checking |\n| | DevOps\u002FSRE | Kelsey Hightower | Automation first, reliability discipline |\n| **Business** | Marketing | Seth Godin | Purple cow, permission marketing, smallest viable audience |\n| | Operations | Paul Graham | Do things that do not scale, ramen profitability |\n| | Sales | Aaron Ross | Predictable revenue, funnel systems |\n| | CFO | Patrick Campbell | Value-based pricing, unit economics |\n| **Intelligence** | Research Analyst | Ben Thompson | Aggregation theory, value chain analysis |\n\nPlus 30+ reusable skills (deep research, scraping, financial modeling, SEO, security audit, UX audit, etc.).\n\n## macOS Quick Start\n\n```bash\n# Prerequisites:\n# - macOS\n# - Codex CLI or Claude Code installed and authenticated\n# - Available model quota\n\n# Clone\ngit clone https:\u002F\u002Fgithub.com\u002FMaxMiksa\u002FAuto-Company.git\ncd auto-company\n\n# Foreground run (live output)\nmake start\n\n# Or install daemon (auto-start + auto-restart)\nmake install\n```\n\n## Windows (WSL) Quick Start\n\n```powershell\n# Prerequisites:\n# - Windows 10\u002F11 + WSL2 (Ubuntu)\n# - Codex CLI or Claude Code installed and authenticated inside WSL\n# - jq and make available inside WSL\n# - Available model quota\n\n# Clone\ngit clone https:\u002F\u002Fgithub.com\u002FMaxMiksa\u002FAuto-Company.git\ncd auto-company\n\n# Start (daemon mode via PowerShell, default engine = claude)\n.\\scripts\\windows\\start-win.ps1\n\n# Switch engine explicitly\n.\\scripts\\windows\\start-win.ps1 -Engine codex\n\n# Status\n.\\scripts\\windows\\status-win.ps1\n\n# Stop\n.\\scripts\\windows\\stop-win.ps1\n```\n\nFor monitoring, dashboard, and autostart commands, see [`docs\u002Fwindows-setup.md`](docs\u002Fwindows-setup.md).\n\n## Command Quick Reference (By Platform)\n\n| Task | macOS \u002F WSL (Terminal) | Windows (PowerShell) |\n|---|---|---|\n| Start | `make start` | `.\\scripts\\windows\\start-win.ps1` |\n| Status | `make status` | `.\\scripts\\windows\\status-win.ps1` |\n| Live logs | `make monitor` | `.\\scripts\\windows\\monitor-win.ps1` |\n| Last cycle output | `make last` | `.\\scripts\\windows\\last-win.ps1` |\n| Cycle summary | `make cycles` | `.\\scripts\\windows\\cycles-win.ps1` |\n| Stop | `make stop` | `.\\scripts\\windows\\stop-win.ps1` |\n| Web dashboard | `make dashboard` | `.\\scripts\\windows\\dashboard-win.ps1` |\n| Install daemon | `make install` | Auto-installed\u002Fstarted by `start-win.ps1` |\n| Uninstall daemon | `make uninstall` | `wsl -d Ubuntu --cd \u003Crepo_wsl_path> bash -lc 'make uninstall'` |\n| Pause daemon | `make pause` | `wsl -d Ubuntu --cd \u003Crepo_wsl_path> bash -lc 'make pause'` |\n| Resume daemon | `make resume` | `wsl -d Ubuntu --cd \u003Crepo_wsl_path> bash -lc 'make resume'` |\n\n### macOS Sleep Prevention (macOS Only)\n\nmacOS screen lock usually does not kill processes, but system sleep can pause work. For long runs:\n\n```bash\nmake start-awake   # Start loop and keep system awake until loop exits\n\n# If loop is already running (after make start):\nmake awake         # Attach caffeinate to PID in .auto-loop.pid\n```\n\nNotes:\n- Both commands depend on built-in `caffeinate`\n- `make awake` exits automatically when target PID exits\n\n## Architecture & Technology Stack (5-Layer Architecture)\n\nAuto-Company is not a simple LLM API wrapper, but a highly decoupled **Multi-Agent System (MAS)**. Its technical architecture is divided into 5 distinct layers:\n\n```text\n┌────────────────────────────────────────────────────────────┐\n│ 5. Observability & HITL (Human-In-The-Loop) Layer          │\n│    [ Dashboard ]  [ File-based Steering (consensus.md) ]   │\n├────────────────────────────────────────────────────────────┤\n│ 4. Workflow Routing & Teaming Layer                        │\n│    [ Dynamic Squad Routing ]  [ Forced Convergence Flow ]  │\n├────────────────────────────────────────────────────────────┤\n│ 3. Agentic Models & Cognition Layer                        │\n│    [ 14 Expert Personas ]  [ 30+ Skill Arsenal ]           │\n├────────────────────────────────────────────────────────────┤\n│ 2. Orchestration & State Machine Layer                     │\n│    [ 24\u002F7 Auto-Loop ]  [ State Machine ]  [ Resilience ]   │\n├────────────────────────────────────────────────────────────┤\n│ 1. Execution Engine & Infrastructure Layer                 │\n│    [ Dual-Engine (Claude\u002FCodex) ]  [ Cross-Platform Daemon]│\n└────────────────────────────────────────────────────────────┘\n```\n\n### Layer 5: Observability & HITL (Human-In-The-Loop)\n*   **File-based Steering**: Humans only need to edit `memories\u002Fconsensus.md` and modify the `Next Action`. The AI team waking up in the next cycle will immediately pivot, enabling minimalist macro-control.\n*   **Full-chain Logs & Dashboard**: `logs\u002F` records the complete output and chain-of-thought for each cycle. `dashboard\u002F` provides a local visualization dashboard based on a Python server, displaying real-time cycle status, cost consumption, and agent activity.\n\n### Layer 4: Workflow Routing & Teaming\n*   **Dynamic Squad Formation**: Powered by Agent Teams, the system dynamically selects 2-5 of the most suitable experts from the 14-person pool based on the \"Next Action\" in `consensus.md`, instantiating them as sub-agents for the current loop.\n*   **Forced Convergence Flow**: Hardcoded flow control in `PROMPT.md`. For example: Cycle 1 Ideation -> Cycle 2 Validation (Pre-mortem, GO\u002FNO-GO) -> Cycle 3 Execution (Code & Deploy, **pure discussion is forbidden**).\n\n### Layer 3: Agentic Models & Cognition\n*   **Expert Personas Injection**: Instead of generic prompts, it injects specific mental models of historical figures\u002Findustry leaders (e.g., Bezos's \"Working Backwards\", Munger's \"Checklists\", DHH's \"Majestic Monolith\") into `.claude\u002Fagents\u002F`, giving decisions extreme business and engineering depth.\n*   **Skill Arsenal**: A pluggable system located in `.claude\u002Fskills\u002F` (e.g., `frontend-design`, `security-audit`). Specific methodologies are encapsulated as tools that any awakened Agent can \"temporarily load\".\n*   **Constitutional Guardrails**: System-level prompts hardcoded in `CLAUDE.md` set absolute bottom lines (e.g., no deleting repos, no force pushes) to ensure safety under high autonomy.\n\n### Layer 2: Orchestration & State Machine\n*   **The Auto-Loop**: The execution loop controlled by `scripts\u002Fcore\u002Fauto-loop.sh` frees the AI from \"single-turn conversations\", enabling 24\u002F7 continuous operation.\n*   **Lightweight State Machine (Consensus Memory)**: Forgoes complex vector databases or memory management, compressing cross-cycle context into a single Markdown file: `memories\u002Fconsensus.md`. Read before every cycle and rewritten before it ends, acting as the system's \"baton\".\n*   **Resilience & Self-Healing**: Built-in circuit breakers (cooldown triggered by consecutive errors), rate-limit backoff (auto-sleep on 429 errors), and sandbox reset (auto-rollback if a valid consensus is not output).\n\n### Layer 1: Execution Engine & Infrastructure\n*   **Dual-Engine Executor**: Acts as the underlying executor by calling mature AI CLIs (**Claude Code** or **Codex CLI**), naturally inheriting their file I\u002FO, Bash execution, and Git operation capabilities.\n*   **Cross-Platform Daemon**: macOS uses `launchd` for auto-start and crash recovery; Windows\u002FWSL runs via `systemd --user` inside a WSL container, controlled and kept alive externally via PowerShell.\n*   **Sandbox Boundary**: Currently relies on underlying CLI configurations (like Codex's `danger-full-access` or Claude's `bypassPermissions`). System-level operations occur directly in the host environment (or WSL container).\n\n## Operating Model\n\n### Automatic Convergence (No Endless Discussion)\n\n| Cycle | Action |\n|------|------|\n| Cycle 1 | Brainstorm: each agent proposes ideas, rank top 3 |\n| Cycle 2 | Validate #1: Munger pre-mortem + Thompson market check + Campbell economics -> **GO \u002F NO-GO** |\n| Cycle 3+ | GO -> create repo, build, deploy. NO-GO -> move to next idea. Discussion-only loops are forbidden |\n\n### Six Standard Workflows\n\n| # | Workflow | Collaboration Chain |\n|---|------|--------|\n| 1 | **New Product Evaluation** | Research -> CEO -> Munger -> Product -> CTO -> CFO |\n| 2 | **Feature Development** | Interaction -> UI -> Full-stack -> QA -> DevOps |\n| 3 | **Product Launch** | QA -> DevOps -> Marketing -> Sales -> Ops -> CEO |\n| 4 | **Pricing and Monetization** | Research -> CFO -> Sales -> Munger -> CEO |\n| 5 | **Weekly Review** | Ops -> Sales -> CFO -> QA -> CEO |\n| 6 | **Opportunity Discovery** | Research -> CEO -> Munger -> CFO |\n\n## Steering\n\nThe team runs autonomously, but you can intervene at any time:\n\n| Method | Action |\n|------|------|\n| **Change direction** | Edit \"Next Action\" in `memories\u002Fconsensus.md` |\n| **Pause** | `make pause` (macOS\u002FWSL daemon mode) or `.\\scripts\\windows\\stop-win.ps1` (Windows entry) |\n| **Resume** | `make resume` |\n| **Review outputs** | Check `docs\u002F*\u002F` for artifacts generated by agents |\n\n## Safety Guardrails\n\nHard constraints in `CLAUDE.md`, enforced for all agents:\n\n- Do not delete GitHub repos (`gh repo delete`)\n- Do not delete Cloudflare projects (`wrangler delete`)\n- Do not delete system directories (`~\u002F.ssh\u002F`, `~\u002F.config\u002F`, etc.)\n- Do not perform illegal activity\n- Do not leak credentials into public repositories\n- Do not force push to main\u002Fmaster\n- Create all new projects under `projects\u002F`\n\n## Configuration\n\nEnvironment variable overrides:\n\n```bash\nENGINE=claude make start                   # Default engine (claude|codex)\nENGINE=codex make start                    # Switch to codex\nMODEL=sonnet make start                    # Optional model override\nCLAUDE_PERMISSION_MODE=bypassPermissions make start  # Claude permission mode\nLOOP_INTERVAL=60 make start                # 60s interval (default 30)\nCYCLE_TIMEOUT_SECONDS=3600 make start      # 1h cycle timeout (default 1800)\nMAX_CONSECUTIVE_ERRORS=3 make start        # Circuit-breaker threshold (default 5)\nCODEX_SANDBOX_MODE=workspace-write make start  # Optional sandbox override\nCLAUDE_BIN=\u002Fusr\u002Flocal\u002Fbin\u002Fclaude make start     # Optional Claude binary override\nCODEX_BIN=\u002Fusr\u002Flocal\u002Fbin\u002Fcodex make start       # Optional Codex binary override\n```\n\nWindows `start-win.ps1` writes the same values into `.auto-loop.env`:\n\n```powershell\n.\\scripts\\windows\\start-win.ps1 -Engine claude -ClaudePermissionMode bypassPermissions\n.\\scripts\\windows\\start-win.ps1 -Engine codex -SandboxMode workspace-write\n# Backward compatibility:\n.\\scripts\\windows\\start-win.ps1 -Engine codex -CodexSandboxMode workspace-write\n```\n\nNo automatic engine fallback is performed. If the selected engine is missing, startup fails fast.\n\n## Project Structure\n\n```\nauto-company\u002F\n├── CLAUDE.md              # Company charter (mission + guardrails + team + workflows)\n├── PROMPT.md              # Per-cycle execution prompt (convergence rules)\n├── Makefile               # Common command entry\n├── INDEX.md               # script index + responsibility table\n├── dashboard\u002F             # Local web status dashboard (macOS: make dashboard, Windows: dashboard-win.ps1)\n├── scripts\u002F\n│   ├── core\u002F              # Core loop and control scripts (auto-loop\u002Fmonitor\u002Fstop)\n│   ├── windows\u002F           # Windows entry\u002Fguardian\u002Fautostart scripts\n│   ├── wsl\u002F               # WSL systemd --user daemon scripts\n│   └── macos\u002F             # macOS launchd daemon scripts\n├── memories\u002F\n│   └── consensus.md       # Shared handoff memory across cycles\n├── docs\u002F                  # Agent outputs (14 folders + Windows guide)\n├── projects\u002F              # Workspace for generated projects\n├── logs\u002F                  # Loop logs\n└── .claude\u002F\n    ├── agents\u002F            # 14 agent definitions (expert personas)\n    ├── skills\u002F            # 30+ reusable skills\n    └── settings.json      # Permissions + Agent Teams switch\n```\n\n## Dependencies\n\n| Dependency | Notes |\n|------|------|\n| **Claude Code \u002F Codex CLI** | Supported CLI engines (default: Claude) |\n| **macOS or Windows + WSL2 (Ubuntu)** | macOS uses launchd; Windows uses WSL execution core |\n| `node` | Runtime for npm-installed CLI tools |\n| `make` | Start\u002Fstop\u002Fmonitor command entry (WSL\u002FmacOS) |\n| `jq` | Recommended for log processing |\n| `gh` | Optional, GitHub CLI |\n| `wrangler` | Optional, Cloudflare CLI |\n\n## FAQ\n\n### 1) WSL `.sh` fails with `^M` \u002F `bad interpreter`\n\n- Cause: CRLF line endings in shell scripts\n- Fix:\n  - Keep LF rules in `.gitattributes`\n  - Run `git config core.autocrlf false && git config core.eol lf`\n\n### 2) WSL says `codex`\u002F`claude` command not found\n\n- Cause: CLI installed on Windows only, missing in WSL\n- Fix: install `node` and your chosen CLI inside WSL (`@openai\u002Fcodex` or Claude Code)\n\n### 3) Claude waits for permission and cycles appear blocked\n\n- Cause: strict permission mode in Claude CLI\n- Fix: set `CLAUDE_PERMISSION_MODE=bypassPermissions` (or pass `-ClaudePermissionMode bypassPermissions` in `start-win.ps1`)\n- Verify: check `logs\u002Fauto-loop.log` for `Engine: claude` and `PermissionMode: ...`\n\n### 4) `make install` fails inside WSL\n\n- Cause: no available `systemctl --user` in current session\n- Fix:\n  - Verify WSL systemd is enabled\n  - Run `systemctl --user --version`\n  - Re-open WSL session and retry if needed\n\n## Disclaimer\n\nThis is an **experimental project**:\n\n- **Daemon mode works on both macOS and WSL**: launchd on macOS, systemd --user on WSL\n- **Windows entry requires WSL**: PowerShell is only the control layer\n- **Still under test**: runs, but stability is not guaranteed\n- **Costs money**: each cycle consumes model quota\n- **Fully autonomous**: agents act without approval prompts; configure guardrails carefully in `CLAUDE.md`\n- **No warranty**: review `docs\u002F` and `projects\u002F` regularly\n\nSuggested rollout: start with `make start` (foreground), then move to daemon mode (`make install` on macOS\u002FWSL, `.\\scripts\\windows\\start-win.ps1` on Windows).\n\n## Acknowledgments\n\n- Thanks to [@JasonQWJ](https:\u002F\u002Fgithub.com\u002FJasonQWJ) and [@cnwillz](https:\u002F\u002Fgithub.com\u002Fcnwillz) for earlier macOS dashboard support proposals and implementation attempts that helped inform the final cross-platform dashboard design released in `v1.1.0`.\n- [nicepkg\u002Fauto-company](https:\u002F\u002Fgithub.com\u002Fnicepkg\u002Fauto-company) - initial macOS edition\n- [continuous-claude](https:\u002F\u002Fgithub.com\u002FAnandChowdhary\u002Fcontinuous-claude) - cross-session shared notes\n- [ralph-claude-code](https:\u002F\u002Fgithub.com\u002Ffrankbria\u002Fralph-claude-code) - exit signal interception\n- [claude-auto-resume](https:\u002F\u002Fgithub.com\u002Fterryso\u002Fclaude-auto-resume) - usage-limit resume pattern\n\n## 🤝 Contribution & Contact\n\nWelcome to submit Issues and Pull Requests!\nAny questions or suggestions? Please contact Zheyuan (Max) Kong (Carnegie Mellon University, Pittsburgh, PA).\n\nZheyuan (Max) Kong: kongzheyuan@outlook.com | zheyuank@tepper.cmu.edu\nGitHub: https:\u002F\u002Fgithub.com\u002FMaxMiksa\u002FAuto-Company\n\r\n",2,"2026-06-11 04:11:22","high_star"]