comgunner/picoclaw-agents

comgunner 资料

最近记录的来源活动: 2026年4月5日 21:39

SkillsMP 收录数据更新: 2026年10月17日 00:09

已收集 skills: 157

GitHub 星标: 7
GitHub Forks: 1

仓库 README

仓库 README 来源：README.md

查看 README 源文件

展开完整 README

<div align="center"> <img src="assets/logo.jpg" alt="PicoClaw" width="512"> <h1>PicoClaw-Agents</h1> <h3>🤖 Multi-Agent Architecture 🚀 Parallel Subagents</h3> [中文](README.zh.md) | [Español](README.es.md) | [Français](README.fr.md) | [日本語](README.ja.md) | [Português](README.pt-br.md) | [Tiếng Việt](README.vi.md) > **Note:** This project is an independent, hobbyist fork of the original [PicoClaw](https://github.com/sipeed/picoclaw) created by **Sipeed**. It is maintained for experimental and educational purposes. All credit for the original core architecture goes to the Sipeed team. | Feature | OpenClaw | NanoBot | PicoClaw | PicoClaw-Agents | | :-------------------- | :------------ | :------------------ | :---------------------------- | :-------------- | | Language | TypeScript | Python | Go | Go | | RAM | >1GB | >100MB | < 10MB | < 45MB | | Startup (0.8GHz core) | >500s | >30s | <1s | <1s | | Cost | Mac Mini 599$ | Most Linux SBC ~50$ | Any Linux Board As low as 10$ | Any Linux | --- ## 📑 Table of Contents - [🤝 Support This Project](#-support-this-project) - [✨ Features](#-features) - [🦾 Demonstration](#-demonstration) - [📦 Quick Start (2 Minutes)](#-quick-start-2-minutes) - [🖥️ Deploy on VPS](#️-deploy-on-vps) - [🔄 Migration](#-migration-from-openclaw-or-nanoclaw) - [🧠 Native Skills](#-native-skills-optional) - [📁 Workspace Skills](#-workspace-skills-custom) - [🔀 Free Models Cascade](#-free-models-cascade-survival-mechanism) - [⚙️ Configuration](#️-configuration) - [📡 Channels](#-channels) - [🛠️ Tools](#️-tools) - [📚 Documentation](#-documentation) - [📝 Changelog](#-changelog) - [📄 License](#-license) --- ## 🤝 Support This Project This project is developed and maintained with the help of AI tools. **🌟 OpenRouter (Free Tier)** — Access 200+ models with generous free tier. Recommended. ```bash ./picoclaw-agents onboard --openrouter # View models: http://localhost:18800/models ``` 🆓 **Free models** (cascade priority): - `openrouter/free` — Auto-selects best free model - `nvidia/nemotron-3-ultra-550b-a55b:free` — 550B parameters - `openai/gpt-oss-20b:free` — Open source GPT - `inclusionai/ling-3.0-tiny:free` — Lightweight, fast - `meta-llama/llama-3.1-8b-instruct:free` — Llama 3.1 8B **Paid fallbacks:** - `deepseek/deepseek-v4-flash-0731` — DeepSeek V4 Flash - `xiaomi/mimo-v2.5` — Xiaomi MiMo V2.5 **Image generation**: `krea/krea-2-medium-turbo` <p align="center"> <img src="assets/robot_arm_cyberpunk.png" alt="Robot Arm Cyberpunk" width="400"> <img src="assets/japanese_garden.png" alt="Japanese Garden" width="400"> </p> > Generated with OpenRouter `krea/krea-2-medium-turbo` via PicoClaw-Agents 👉 [Join now](https://openrouter.ai/) **🌟 OpenCode Zen (8 Free Models)** — No credit card needed. ```bash ./picoclaw-agents onboard --opencode ``` 🆓 **Free models**: - `opencode/mimo-v2.5-free` — Xiaomi MiMo V2.5 - `opencode/deepseek-v4-flash-free` — DeepSeek V4 Flash - `opencode/nemotron-3-ultra-free` — NVIDIA 550B - `opencode/ling-3.0-tiny-free` — Lightweight, fast - `opencode/north-mini-code-free` — Code-focused - `opencode/laguna-s-2.1-free` — New - `opencode/longcat-2.0-free` — New - `opencode/big-pickle` — Free trial 👉 [Get API Key](https://opencode.ai/auth) **🌟 Zhipu AI (z.ai)** — 100% FREE with `glm-4.5-flash` ```bash ./picoclaw-agents onboard --zhipu ``` 🚀 You're invited to join the GLM Coding Plan. Full support for Claude Code, Cline, and 20+ top coding tools — just $10/month. 👉 [Join now](https://z.ai/subscribe?ic=RF2YMCHBHL) **🌟 Qwen Code (Alibaba Cloud)** — AI coding assistant (paid plans available). ```bash ./picoclaw-agents onboard --qwen # View models: http://localhost:18800/models ``` 👉 [Get started](https://www.alibabacloud.com/campaign/benefits?referral_code=A924LX) > ℹ️ **Note:** Use `onboard` for first-time setup (configures everything automatically). Use `auth` only if you need to add another API key to an existing configuration. > > **Example:** You already ran `onboard --openrouter` and now want to add OpenCode Zen: > ```bash > # First time (sets up OpenRouter as default) > ./picoclaw-agents onboard --openrouter > > # Later, add OpenCode Zen as additional provider > ./picoclaw-agents auth login --provider opencode > > # Now you can switch between them > /model opencode/mimo-v2.5-free # Use OpenCode model > /model openrouter/free # Back to OpenRouter > ``` > > 💡 Using these links helps support the ongoing development of PicoClaw-Agents. Thank you! ## ✨ Features * 🪶 **Ultra-Lightweight**: Optimized Go implementation with minimal footprint. * 🤖 **Multi-Agent Architecture**: Fail-Close security (detects invalid config), robust channel/bus closure handling, and Skills Sentinel (native security layer) with proactive input/output sanitization and local auditing (`AUDIT.md`). * 🚀 **Parallel Subagents**: Spawn multiple autonomous subagents working in parallel, each with independent model configurations. * 🌍 **True Portability**: Single self-contained binary across RISC-V, ARM, and x86. * 🦾 **AI-Bootstrapped**: Core implementation refined through autonomous agentic workflows. * 📈 **Binance Integration**: Native trading tools for spot balances, futures positions (LONG/SHORT), and public ticker data via direct API or MCP server. * 📱 **Social Media Tools**: Publish to Facebook (images + messages) and X/Twitter (tweets + threads) with multi-page support and automatic comment fallback. * 🎨 **AI Image Generation**: Generate images from text prompts using **Antigravity OAuth** (`gemini-3.1-flash-image`, no API key needed, 300s cooldown), **Gemini API key**, or **Ideogram API key**. Includes script-to-image workflow and auto-posting to social media. * 📝 **Notion Integration**: Create, query, update pages and databases for task tracking and data management. * 🤖 **Community Manager**: Auto-generate engaging social media posts from technical content or generated images. * ⚡ **Fast-path Slash Commands**: Instant system commands via `/` or `#` that bypass the LLM for zero-latency approvals, status checks, and lotes (bundle) management. Works on Telegram, Discord, and CLI. * 🖇️ **Global State Synchronization**: Global ImageGenTracker, a shared memory space across all agents (PM, Subagents) to ensure perfect consistency in multi-agent workflows. * 🚀 **Native Skills Architecture**: Native skills compiled directly into the binary (`pkg/skills/queue_batch.go`), eliminating external file dependencies and enhancing security. See [docs/QUEUE_BATCH.en.md](docs/QUEUE_BATCH.en.md). <img src="assets/compare.jpg" alt="PicoClaw" width="512"> ## 🦾 Demonstration ### 🛠️ Standard Assistant Workflows <table align="center"> <tr align="center"> <th><p align="center">🧩 Full-Stack Engineer</p></th> <th><p align="center">🗂️ Logging & Planning Management</p></th> <th><p align="center">🔎 Web Search & Learning</p></th> <th><p align="center">🔧 General Worker</p></th> </tr> <tr> <td align="center"></td> <td align="center"></td> <td align="center"></td> <td align="center"></td> </tr> <tr> <td align="center">Develop • Deploy • Scale</td> <td align="center">Schedule • Automate • Memory</td> <td align="center">Discovery • Insights • Trends</td> <td align="center">Tasks • Support • Efficiency</td> </tr> </table> ### 🚀 Advanced Multi-Agent Workflow (The "Dream Team") Take advantage of the subagent architecture to deploy a full software development lifecycle team. **The "DevOps & QA" Team (Powered by [DeepSeek Reasoner](https://platform.deepseek.com)):** * **`project_manager` (Leader)**: Has permission to spawn any agent. Oversees the global objective and delegates sub-tasks. * **`senior_dev` (The Engine)**: Technical expert. Spawns the QA Specialist to review code or the Junior Fixer for boilerplate. * **`qa_specialist` (Ops & Testing)**: Quality logic. Tests code, finds bugs, proposes fixes, and manages GitHub deployments. * **`junior_fixer` (The Assistant)**: Focuses on small fixes, refactoring, and documentation under supervision. * **`general_worker` (The Groundwork)**: Versatile agent for common tasks, information retrieval, and supporting the rest of the team. **How to use this?** Simply send a high-level command to the Leader via Telegram or CLI: > *"Leader, I need the Senior Dev to fix the database bug and have the QA specialist verify the build before pushing to GitHub."* PicoClaw will automatically manage the hierarchy: **PM ➔ Senior Dev ➔ QA Specialist (Fix & Publish).** > [!TIP] > **Check out the examples:** See `config_dev.example.json` for a standard DeepSeek team, `config_dev_multiple_models.example.json` for a mixed-model team (OpenAI, Anthropic, and DeepSeek), and `config_context_management.example.json` for optimizing token usage during extensive coding sessions. --- ### 🛠️ MCP Builder Agent - Build Custom AI Tools **MCP Builder Agent** (`specialized-mcp-builder`) is a specialized skill for building **Model Context Protocol (MCP)** servers. Create custom tools that extend your AI agent's capabilities — from API integrations to database access to workflow automation. #### What Can You Build? - 🔌 **API Integrations**: Connect to Stripe, GitHub, Slack, or any REST API - 🗄️ **Database Access**: Safely expose data from PostgreSQL, MySQL, MongoDB - 📁 **File Operations**: Controlled read/write access to file systems - 🔄 **Workflow Automation**: Automate business processes (invoices, orders, notifications) - 🎯 **Custom Tools**: Any tool your imagination can design! #### Quick Example: GitHub MCP Server **1. Invoke MCP Builder:** ```bash picoclaw-agents agent -m "Build an MCP server for GitHub API with tools to: search repos, get commits, create issues" ``` **2. Generated Code:** ```typescript // github-server.ts import { McpServer } from "@modelcontextprotocol/sdk/server/mcp.js"; import { StdioServerTransport } from "@modelcontextprotocol/sdk/server/stdio.js"; import { z } from "zod"; import { Octokit } from "octokit"; const server = new McpServer({ name: "github-server", version: "1.0.0" }); const octokit = new Octokit({ auth: process.env.GITHUB_TOKEN }); server.tool( "search_repositories", { query: z.string().describe("Search query"), per_page: z.number().default(10) }, async ({ query, per_page }) => { const { data } = await octokit.request('GET /search/repositories', { q: query, per_page: Math.min(per_page, 100) }); return { content: [{ type: "text", text: JSON.stringify(data.items.map(repo => ({ name: repo.full_name, stars: repo.stargazers_count, url: repo.html_url })), null, 2) }] }; } ); const transport = new StdioServerTransport(); await server.connect(transport); ``` **3. Configure in PicoClaw:** ```json { "tools": { "mcp": { "github": { "command": "node", "args": ["/path/to/github-server.ts"], "env": { "GITHUB_TOKEN": "ghp_..." } } } } } ``` **4. Use with Your Agent:** ``` @agent Search for machine learning repositories with more than 1000 stars ``` **Output:** ``` Found 15 repositories: - tensorflow/tensorflow (178k stars) - pytorch/pytorch (85k stars) - scikit-learn/scikit-learn (58k stars) ... ``` #### Best Practices ✅ **DO:** - Use descriptive names: `search_users_by_email` not `query1` - Validate all inputs with Zod schemas - Provide detailed parameter descriptions - Handle errors gracefully with actionable messages - Write unit tests for each tool ❌ **DON'T:** - Use generic names like `tool1`, `do_stuff` - Skip input validation - Return cryptic error messages - Rely on state between calls (keep tools stateless) #### Documentation 📖 **Complete Guide:** See [docs/MCP_BUILDER_AGENT.md](docs/MCP_BUILDER_AGENT.md) for: - Full API reference - Step-by-step examples (GitHub, PostgreSQL, Stripe) - Best practices and patterns - Testing strategies #### Available Skills MCP Builder is one of **158+ specialized skills** available in PicoClaw: - **Specialized**: `mcp-builder`, `salesforce-architect`, `blockchain-security-auditor` - **Engineering**: `backend-architect`, `devops-automator`, `security-engineer` - **Marketing**: `seo-specialist`, `social-media-strategist`, `content-creator` - **Support**: `legal-compliance-checker`, `finance-tracker`, `infrastructure-maintainer` See full list in [CHANGELOG.md](CHANGELOG.md#370---2026-03-26). ### 📱 Run on old Android Phones Give your decade-old phone a second life! Turn it into a smart AI Assistant with PicoClaw. Quick Start: 1. **Install Termux** (Available on F-Droid or Google Play). 2. **Execute cmds** ```bash # Note: Replace v0.1.1 with the latest version from the Releases page wget https://github.com/comgunner/picoclaw-agents/releases/download/v0.1.1/picoclaw-agents_Linux_arm64 chmod +x picoclaw-agents_Linux_arm64 pkg install proot termux-chroot ./picoclaw-agents_Linux_arm64 onboard ``` And then follow the instructions in the "Quick Start" section to complete the configuration! <img src="assets/termux.jpg" alt="PicoClaw" width="512"> ### 🐜 Innovative Low-Footprint Deploy PicoClaw can be deployed on almost any Linux device, from simple embedded boards to powerful servers. ## 🚀 Launchers PicoClaw-Agents includes two optional graphical launchers for users who prefer a visual interface. ### 💻 TUI Launcher (Recommended for Headless / SSH) The TUI (Terminal UI) Launcher provides a full-featured terminal interface for configuration and management. Ideal for servers, Raspberry Pi, and headless environments. **Build:** ```bash make build-launcher-tui ``` **Run:** ```bash picoclaw-agents-launcher-tui # Or in development mode make dev-launcher-tui ``` **Features:** - Interactive terminal menu (arrow keys + hotkeys) - Model configuration - Channel management (Telegram, Discord, etc.) - Gateway control (start/stop daemon) - Interactive chat with AI - TOML-based configuration ![TUI Launcher](assets/launcher-tui.jpg) --- ### 🌐 WebUI Launcher The WebUI Launcher provides a browser-based interface for configuration and chat. No command-line knowledge required. **Build Frontend:** ```bash cd web/frontend pnpm install pnpm build:backend # Assets in: web/backend/dist/ ``` **Features:** - Browser-based configuration UI - Visual channel management - Gateway control panel - Session history viewer - Skills management - Model configuration - Multi-language support (English, 简体中文，Español) **Usage:** ```bash make build-launcher picoclaw-agents-launcher # Open http://localhost:18800 in your browser ``` > **Tip — Remote access / Docker / VM**: Add the `-public` flag to listen on all interfaces: > ```bash > picoclaw-agents-launcher -public > ``` **OAuth Authentication via Web UI:** You can authenticate with OAuth providers directly from the Web UI at `http://localhost:18800/credentials`: - **Anthropic**: Browser OAuth (PKCE flow) — Auto-adds 5 Claude models - **Google Antigravity**: Browser OAuth — Auto-adds 15 Gemini models - **OpenAI**: Device Code only — Auto-adds 8 GPT models - **Qwen Portal**: API Key (token) — Auto-adds 6 Qwen models - **Zhipu AI (z.ai)**: API Key (token) — Auto-adds 7 GLM models 🆓 **100% FREE** with `glm-4.5-flash` **CLI Authentication Commands:** ```bash # OpenAI (Codex OAuth - Device Code) ./picoclaw-agents auth login --provider openai --device-code # Google Antigravity (OAuth) ./picoclaw-agents auth login --provider google-antigravity # Qwen Portal (DashScope) ./picoclaw-agents auth login --provider qwen # Zhipu AI (z.ai) - 100% FREE with glm-4.5-flash ./picoclaw-agents auth login --provider zhipu # DeepSeek AI ./picoclaw-agents auth login --provider deepseek # Kilo AI ./picoclaw-agents auth login --provider kilo ``` All methods support API key token input or OAuth flow and automatically configure models on successful authentication. ![Credentials OAuth](assets/webui/credentials-auth.png) > **Note:** OpenAI only supports **Device Code** authentication (no Browser OAuth). Use `--device-code` flag or the Web UI Device Code button. ![WebUI Launcher](assets/launcher-webui.jpg) --- ## 🖥️ OS Service Management Run PicoClaw-Agents as a system service that starts automatically on login. Supports **Linux (systemd)**, **macOS (launchd)**, and **Windows (Scheduled Tasks)**. > **Note:** The `service` subcommand is in the **main CLI binary**, not the launcher. ```bash # ✅ Correct — main CLI binary ./build/picoclaw-agents-darwin-arm64 service install # ❌ Incorrect — launcher is WebUI only ./build/picoclaw-agents-launcher-darwin-arm64 service install # Error ``` ### Available Commands | Command | Description | |---------|-------------| | `service install` | Install as OS service (auto-start on login) | | `service install --dry-run` | Preview what would be installed | | `service install --port 9999` | Install on custom port | | `service install --public` | Listen on all interfaces (0.0.0.0) | | `service uninstall` | Remove the service | | `service start` | Start the service | | `service stop` | Stop the service | | `service restart` | Restart the service | | `service status` | Check service status | | `service logs` | View logs (last 50 lines) | | `service logs -n 100` | View last 100 lines | | `service logs -f` | Follow logs in real-time | ### Platform Details | Platform | Service Type | Config Location | |----------|-------------|-----------------| | **Linux** | systemd user unit | `~/.config/systemd/user/picoclaw-agents.service` | | **macOS** | launchd LaunchAgent | `~/Library/LaunchAgents/com.picoclaw.agents.gateway.plist` | | **Windows** | Scheduled Task (ONLOGON) | `%APPDATA%\PicoClaw\gateway.cmd` | ### Typical Workflow ```bash # 1. Install the service (auto-starts on login) picoclaw-agents service install # 2. Open the WebUI for visual configuration picoclaw-agents-launcher --public # → http://localhost:18800 # 3. The service manages the gateway automatically # No need to launch it manually ``` ### ⚠️ Prevent Duplicate Connections The service and a manually-started gateway **must not run simultaneously**. If both are active, you'll get duplicate messages on Telegram/Discord and port conflicts. The installer checks if the gateway port is already in use and refuses to install if it detects a running gateway. ``` Error: gateway is already running on port 18800. Stop it first to avoid duplicate connections. ``` ### Dry Run Example ```bash $ picoclaw-agents service install --dry-run 🔍 Dry Run Mode — No changes will be made ================================================== Platform: macOS (launchd) Port: 18800 Public: false Config: ~/.picoclaw/config.json Pre-flight checks: ✓ Port 18800 is available ✓ Service is not already installed ✓ Binary exists What would be done: 1. Create directory: ~/Library/Logs/picoclaw-agents 2. Write plist file: ~/Library/LaunchAgents/com.picoclaw.agents.gateway.plist 3. Run: launchctl bootstrap gui/$(id -u)/ ... ``` 📖 **Full documentation:** See [docs/SERVICE.md](docs/SERVICE.md) for installation guides, troubleshooting, and platform-specific details. --- ## 📦 Install ### Install with precompiled binary #### 🍎 macOS (Apple Silicon - M1/M2/M3) **Direct download and install:** ```bash # Download latest release curl -LO https://github.com/comgunner/picoclaw-agents/releases/latest/download/picoclaw-agents_Darwin_arm64.tar.gz # Extract tar -xzf picoclaw-agents_Darwin_arm64.tar.gz # Make executable chmod +x picoclaw-agents # Move to PATH (optional) sudo mv picoclaw-agents /usr/local/bin/ # Verify installation picoclaw-agents --version ``` #### 🍎 macOS (Intel - x86_64) ```bash curl -LO https://github.com/comgunner/picoclaw-agents/releases/latest/download/picoclaw-agents_Darwin_x86_64.tar.gz tar -xzf picoclaw-agents_Darwin_x86_64.tar.gz chmod +x picoclaw-agents sudo mv picoclaw-agents /usr/local/bin/ ``` #### 🪟 Windows (x86_64) **PowerShell (Admin):** ```powershell # Download latest release Invoke-WebRequest -Uri "https://github.com/comgunner/picoclaw-agents/releases/latest/download/picoclaw-agents_Windows_x86_64.zip" -OutFile "picoclaw-agents.zip" # Extract Expand-Archive -Path "picoclaw-agents.zip" -DestinationPath "$env:USERPROFILE\picoclaw-agents" # Add to PATH (optional - requires admin) $env:Path += ";$env:USERPROFILE\picoclaw-agents" [Environment]::SetEnvironmentVariable("Path", $env:Path, "User") # Verify picoclaw-agents --version ``` #### 🐧 Linux ```bash # ARM64 (Raspberry Pi 4, AWS Graviton, etc.) curl -LO https://github.com/comgunner/picoclaw-agents/releases/latest/download/picoclaw-agents_Linux_arm64.tar.gz tar -xzf picoclaw-agents_Linux_arm64.tar.gz chmod +x picoclaw-agents sudo mv picoclaw-agents /usr/local/bin/ # x86_64 (Intel/AMD) curl -LO https://github.com/comgunner/picoclaw-agents/releases/latest/download/picoclaw-agents_Linux_x86_64.tar.gz tar -xzf picoclaw-agents_Linux_x86_64.tar.gz chmod +x picoclaw-agents sudo mv picoclaw-agents /usr/local/bin/ ``` #### 📦 All Platforms Download the firmware for your platform from the [releases page](https://github.com/comgunner/picoclaw-agents/releases). | Platform | Architecture | File | |----------|--------------|------| | macOS | Apple Silicon (M1/M2/M3) | `picoclaw-agents_Darwin_arm64.tar.gz` | | macOS | Intel (x86_64) | `picoclaw-agents_Darwin_x86_64.tar.gz` | | Linux | ARM64 | `picoclaw-agents_Linux_arm64.tar.gz` | | Linux | x86_64 | `picoclaw-agents_Linux_x86_64.tar.gz` | | Linux | ARMv7 | `picoclaw-agents_Linux_armv7.tar.gz` | | Windows | x86_64 | `picoclaw-agents_Windows_x86_64.zip` | | Windows | ARM64 | `picoclaw-agents_Windows_arm64.zip` | ### Install from source (latest features, recommended for development) ```bash git clone https://github.com/comgunner/picoclaw-agents.git cd picoclaw-agents make deps # Build, no need to install make build # Build for multiple platforms make build-all # Build And Install make install ``` ## 🐳 Docker Compose You can also run PicoClaw using Docker Compose without installing anything locally. ```bash # 1. Clone this repo git clone https://github.com/comgunner/picoclaw-agents.git cd picoclaw-agents # 2. Set your API keys cp config/config.example.json config/config.json vim config/config.json # Set DISCORD_BOT_TOKEN, API keys, etc. # 3. Build & Start docker compose --profile gateway up -d > [!TIP] > **Docker Users**: By default, the Gateway listens on `127.0.0.1` which is not accessible from the host. If you need to access the health endpoints or expose ports, set `PICOCLAW_GATEWAY_HOST=0.0.0.0` in your environment or update `config.json`. # 4. Check logs docker compose logs -f picoclaw-gateway # 5. Stop docker compose --profile gateway down ``` ### Agent Mode (One-shot) ```bash # Ask a question docker compose run --rm picoclaw-agents-agent -m "What is 2+2?" # Interactive mode docker compose run --rm picoclaw-agents-agent ``` ### Rebuild ```bash docker compose --profile gateway build --no-cache docker compose --profile gateway up -d ``` ### 🚀 Quick Start > [!TIP] > Set your API key in `~/.picoclaw/config.json`. > Get API keys: [OpenRouter](https://openrouter.ai/keys) (LLM) · [Zhipu](https://open.bigmodel.cn/usercenter/proj-mgmt/apikeys) (LLM) > Web Search is **optional** - get free [Tavily API](https://tavily.com) (1000 free queries/month) or [Brave Search API](https://brave.com/search/api) (2000 free queries/month) or use built-in auto fallback. **1. Initialize** Use the `onboard` command to initialize your workspace with a pre-configured template for your preferred provider: ```bash # Default (Empty/Manual configuration) picoclaw-agents onboard # 🆓 Zero-cost setup — no API balance required: picoclaw-agents onboard --free # Free tier (OpenRouter free models) # Pre-configured templates: picoclaw-agents onboard --openrouter # Use OpenRouter template (openrouter/free, default) picoclaw-agents onboard --opencode # Use OpenCode Zen template (8 free models) picoclaw-agents onboard --openai # Use OpenAI template (o3-mini) picoclaw-agents onboard --glm # Use GLM-4.5-Flash template (zhipu.ai) picoclaw-agents onboard --qwen # Use Qwen template (Alibaba Cloud Intl) picoclaw-agents onboard --qwen_zh # Use Qwen template (Alibaba Cloud China) picoclaw-agents onboard --gemini # Use Gemini template (gemini-2.5-flash) ``` > [!TIP] > **No API key balance?** Use `picoclaw-agents onboard --free` to get started instantly with OpenRouter's free-tier models. Just create a free account at [openrouter.ai](https://openrouter.ai) and add your key — no credits needed. #### 🤖 Team Mode During the onboard wizard, you can now choose **Team Mode** to deploy multiple specialized agents: **Available team templates:** | Template | Agents | Description | |----------|--------|-------------| | **Dev Team** | 9 agents | Engineering Manager + 8 specialists (backend, frontend, devops, qa, security, data, ml, researcher) | | **Research Team** | 3 agents | Coordinator + Researcher + Data Analyst | | **General Team** | 3 agents | Orchestrator + 2 general workers | | **Solo Agent** | 1 agent | Single general-purpose agent (default) | **14 Native Skills available:** | Category | Skills | |----------|--------| | **General** | `fullstack_developer`, `agent_team_workflow`, `researcher` | | **Development** | `backend_developer`, `frontend_developer`, `devops_engineer`, `qa_engineer`, `security_engineer` | | **Data & AI** | `data_engineer`, `ml_engineer` | | **Automation** | `n8n_workflow`, `queue_batch`, `binance_mcp`, `odoo_developer` | **Solo Mode:** Select individual skills to enable for your single agent. **Team Mode:** Pre-configured agents with role-specific skills and subagent spawning rules. #### 🧠 Context Window Management **Tool Result Pruning:** Automatically truncates large tool outputs before sending to the LLM, reducing token usage by ~60%. ```json { "context_management": { "pruning": { "enabled": true, "max_tool_result_chars": 8000, "exclude_tools": ["memory_store", "memory_read"], "aggressive_tools": ["shell", "web_fetch"] } } } ``` **Advanced Compaction:** Configure a separate model for context compaction (uses same provider): ```json { "context_management": { "compaction": { "model": "claude-haiku-4-5-20251001", // Same provider, different model "max_summary_tokens": 2048, // 4x more context preserved "recent_turns_preserve": 6 // Keep last 6 turns verbatim } } } ``` **Manual Compaction Command:** Force immediate context compaction: ```bash # Basic compaction /compact # With focus instructions /compact focus on API changes /compact summarize database operations ``` #### 🆓 Free Tier Models The `--free` flag configures three OpenRouter free-tier models with automatic fallback: | Priority | Model | Context | Notes | |----------|-------|---------|-------| | Primary | `openrouter/free` | varies | Auto-selects best available free model **(default)** | | Fallback 1 | `stepfun/step-3.5-flash` | 256K | High-context reasoning tasks | | Fallback 2 | `deepseek/deepseek-v3.2-20251201` | 64K | Fast, reliable fallback | All three are routed through [OpenRouter](https://openrouter.ai) — a single API key covers all of them. > [!IMPORTANT] > **Model ID Fix:** Earlier versions used `openrouter/free` which was not a valid OpenRouter model ID. This has been fixed — `openrouter/free` is now the recommended default for new installations. If you have an existing config with `openrouter-free`, update it to `openrouter/free` or re-run `picoclaw-agents onboard --free`. > [!TIP] > **OpenAI OAuth on Free Tier:** You can also use OpenAI OAuth authentication (`picoclaw-agents auth login --provider openai --device-code`) which works with free tier plans. No API key required — uses your existing OpenAI/ChatGPT account. **Learn more:** See [docs/OPENROUTER_FREE.md](docs/OPENROUTER_FREE.md) for complete setup guide, rate limits, and troubleshooting. **2. Configure** (`~/.picoclaw/config.json`) ```json { "agents": { "defaults": { "workspace": "~/.picoclaw/workspace", "model_name": "deepseek-chat", "max_tokens": 8192, "temperature": 0.7, "max_tool_iterations": 20, "subagents": { "max_spawn_depth": 2, "max_children_per_agent": 5 } }, "backend_coder": { "model_name": "deepseek-reasoner", "temperature": 0.2 } }, "model_list": [ { "model_name": "deepseek-chat", "model": "deepseek/deepseek-chat", "api_key": "your-api-key" }, { "model_name": "deepseek-reasoner", "model": "deepseek/deepseek-reasoner", "api_key": "your-api-key" } ], "tools": { "web": { "brave": { "enabled": false, "api_key": "YOUR_BRAVE_API_KEY", "max_results": 5 }, "tavily": { "enabled": false, "api_key": "YOUR_TAVILY_API_KEY", "max_results": 5 }, "duckduckgo": { "enabled": true, "max_results": 5 } } } } ``` > **New in v3 (Multi-Agent Architecture)**: You can now spin up isolated **Subagents** to perform parallel background tasks. Crucially, **each subagent can use a completely different LLM model**. As shown in the configuration above, the main agent runs `gpt4`, but it can spawn a dedicated `coder` subagent running `claude-sonnet-4.6` to handle complex programming tasks simultaneously! > **New**: The `model_list` configuration format allows zero-code provider addition. See [Model Configuration](#model-configuration-model_list) for details. > `request_timeout` is optional and uses seconds. If omitted or set to `<= 0`, PicoClaw uses the default timeout (120s). **3. Get API Keys** * **LLM Provider**: [DeepSeek](https://platform.deepseek.com) (Recommended) · [OpenRouter](https://openrouter.ai/keys) · [Zhipu](https://open.bigmodel.cn/usercenter/proj-mgmt/apikeys) · [Anthropic](https://console.anthropic.com) · [OpenAI](https://platform.openai.com) · [Gemini](https://aistudio.google.com/api-keys) * **Web Search** (optional): [Tavily](https://tavily.com) - Optimized for AI Agents (1000 requests/month) · [Brave Search](https://brave.com/search/api) - Free tier available (2000 requests/month) ### 💡 Recommended Models for Developers (`backend_coder`) For heavy coding tasks, performance and logic are key. We recommend standardizing on these models for your `backend_coder` agents: * **DeepSeek**: `deepseek-reasoner` (Excellent reasoning and cost-effective) * **OpenAI**: `o3-mini-2025-01-31` (High performance) * **OpenRouter.ai**: `Qwen3 Coder Plus`, `GPT-5.3-Codex` (Great coding versatility) * **Anthropic**: `Claude Haiku 4.5` (Fast and reliable) > **Note**: See `config.example.json` for a complete configuration template. ### 🧠 Native Skills (Optional) Native skills inject specialized AI personas directly into the agent's system prompt. When enabled, the agent "becomes" that role — no external files, everything compiled into the binary. **Enable in `~/.picoclaw/config.json`:** ```json { "agents": { "defaults": { "skills": ["backend_developer", "researcher"] } } } ``` **All 13 available native skills:** | Skill | Description | |-------|-------------| | `queue_batch` | Batch processing and queue management | | `agent_team_workflow` | Orchestrates multi-agent team workflows | | `fullstack_developer` | Full-stack web development (frontend + backend) | | `n8n_workflow` | n8n automation workflow design | | `binance_mcp` | Binance trading via MCP protocol | | `researcher` | Deep research, analysis, and synthesis | | `backend_developer` | REST APIs, databases, microservices | | `frontend_developer` | React, Vue, CSS, UX patterns | | `devops_engineer` | CI/CD, Docker, Kubernetes, IaC | | `security_engineer` | Security reviews, threat modeling, hardening | | `qa_engineer` | Testing strategies, automation, quality | | `data_engineer` | Pipelines, ETL, data warehousing | | `ml_engineer` | ML/AI model development and deployment | > **Skills vs Tools:** Skills inject context into the system prompt (the agent *becomes* the role). Tools are callable actions (functions the LLM can invoke). Configure them separately: `"skills"` for roles, `"tools_override"` for callable tools. See [`docs/SKILLS.md`](docs/SKILLS.md) for details. ### 📁 Workspace Skills (Custom) Workspace skills are file-based skills stored in `~/.picoclaw/workspace/skills/`. They are automatically discovered by the gateway. **Available workspace skills:** | Skill | Description | |-------|-------------| | `ui-ux-pro-max` | UI/UX design intelligence: 84 styles, 192 color palettes, 74 font pairings, 22 stacks | **Install custom skills:** ```bash # Create skill directory mkdir -p ~/.picoclaw/workspace/skills/my-skill # Create SKILL.md with frontmatter cat > ~/.picoclaw/workspace/skills/my-skill/SKILL.md << 'EOF' --- name: my-skill description: "My custom skill description" --- # Skill content here EOF # Restart gateway to discover skill sudo systemctl restart picoclaw.service ``` See [`local_work/HOW_TO_ADD_CUSTOM_SKILLS.md`](local_work/HOW_TO_ADD_CUSTOM_SKILLS.md) for detailed instructions. ### 🔀 Free Models Cascade (Survival Mechanism) PicoClaw uses a **cascade fallback system** to keep your bot online 24/7. When the primary free model fails, it automatically tries the next model. **Free models (no credit card required):** | Priority | Model | Provider | |----------|-------|----------| | 1 | `openrouter/free` | OpenRouter (auto-selects best available) | | 2 | `nvidia/nemotron-3-ultra-550b-a55b:free` | NVIDIA | | 3 | `openai/gpt-oss-20b:free` | OpenAI | | 4 | `inclusionai/ling-3.0-tiny:free` | InclusionAI | | 5 | `meta-llama/llama-3.1-8b-instruct:free` | Meta | **Paid fallbacks (optional):** | Model | Provider | Cost | |-------|----------|------| | `deepseek/deepseek-v4-flash-0731` | DeepSeek | $0.27/M tokens | | `xiaomi/mimo-v2.5` | Xiaomi | $0.10/M tokens | See [`docs/FREE_MODELS_CASCADE_GUIDE.md`](docs/FREE_MODELS_CASCADE_GUIDE.md) for configuration and troubleshooting. **4. Chat** ```bash picoclaw-agents agent -m "What is 2+2?" ``` That's it! You have a working AI assistant in 2 minutes. --- ## 🖥️ Deploy on VPS Run PicoClaw-Agents on a cloud VPS for 24/7 availability. Recommended specs: | Spec | Value | |------|-------| | **Provider** | [Vultr](https://www.vultr.com/?ref=9916942) | | **Instance** | vc2-1c-2gb | | **CPU** | 1 vCPU | | **RAM** | 2 GB | | **Storage** | 55 GB | | **Price** | From $10/month | > 💡 [**Deploy on Vultr**](https://www.vultr.com/?ref=9916942) — Sufficient for PicoClaw-Agents with OpenRouter free models. ```bash # Install on Ubuntu/Debian curl -fsSL https://raw.githubusercontent.com/comgunner/picoclaw-agents/main/scripts/install.sh | bash # Or manual wget https://github.com/comgunner/picoclaw-agents/releases/latest/download/picoclaw-agents-linux-amd64 chmod +x picoclaw-agents-linux-amd64 sudo mv picoclaw-agents-linux-amd64 /usr/local/bin/picoclaw-agents ``` --- ## 🔄 Migration from OpenClaw or NanoClaw If you're migrating from **OpenClaw** or **NanoClaw** to PicoClaw-Agents, use the `migrate` command: ```bash # Migrate from OpenClaw (default) picoclaw-agents migrate # Explicit OpenClaw migration picoclaw-agents migrate --from openclaw # Migrate from NanoClaw (~/.nanoclaw or ~/.config/nanoclaw) picoclaw-agents migrate --from nanoclaw # Dry-run (preview changes without applying) picoclaw-agents migrate --from nanoclaw --dry-run # Show JSON config diff in dry-run mode picoclaw-agents migrate --from nanoclaw --dry-run --show-diff # Custom NanoClaw home directory picoclaw-agents migrate --from nanoclaw --nanoclaw-home /path/to/nanoclaw # Custom PicoClaw home directory picoclaw-agents migrate --from nanoclaw --picoclaw-home /path/to/picoclaw # Force migration without confirmation picoclaw-agents migrate --from nanoclaw --force ``` **What gets migrated:** | NanoClaw/OpenClaw | → | PicoClaw-Agents | |-------------------|---|-----------------| | `providers[].apiKey` | → | `providers.*.api_key` | | `agents[].model` | → | `agents.defaults.model_name` | | `channels[].telegram.token` | → | `channels.telegram.token` | | `groups/default/CLAUDE.md` | → | `workspace/AGENTS.md` | | `memory/` | → | `workspace/memory/` | | `skills/` | → | `workspace/skills/` | **All migrate flags:** | Flag | Description | |------|-------------| | `--from openclaw\|nanoclaw` | Source to migrate from (default: openclaw) | | `--dry-run` | Show what would be migrated without making changes | | `--show-diff` | Show JSON config diff in dry-run mode | | `--force` | Skip confirmation prompts | | `--config-only` | Only migrate config, skip workspace files | | `--workspace-only` | Only migrate workspace files, skip config | | `--refresh` | Re-sync workspace files from source | | `--nanoclaw-home` | Override NanoClaw home directory | | `--openclaw-home` | Override OpenClaw home directory | | `--picoclaw-home` | Override PicoClaw home directory | --- ## 💬 Chat Apps Talk to your picoclaw-agents through Telegram, Discord, DingTalk, LINE, or WeCom | Channel | Setup | | ------------ | ---------------------------------- | | **Telegram** | Easy (just a token) | | **Discord** | Easy (bot token + intents) | | **QQ** | Easy (AppID + AppSecret) | | **DingTalk** | Medium (app credentials) | | **LINE** | Medium (credentials + webhook URL) | | **WeCom** | Medium (CorpID + webhook setup) | <details> <summary><b>Telegram</b> (Recommended)</summary> **1. Create a bot** * Open Telegram, search `@BotFather` * Send `/newbot`, follow prompts * Copy the token **2. Configure** ```json { "channels": { "telegram": { "enabled": true, "token": "YOUR_BOT_TOKEN", "allow_from": ["YOUR_USER_ID"] } } } ``` > Get your user ID from `@userinfobot` on Telegram. **3. Run** ```bash picoclaw-agents gateway ``` </details> <details> <summary><b>Discord</b></summary> **1. Create a bot** * Go to <https://discord.com/developers/applications> * Create an application → Bot → Add Bot * Copy the bot token **2. Enable intents** * In the Bot settings, enable **MESSAGE CONTENT INTENT** * (Optional) Enable **SERVER MEMBERS INTENT** if you plan to use allow lists based on member data **3. Get your User ID** * Discord Settings → Advanced → enable **Developer Mode** * Right-click your avatar → **Copy User ID** **4. Configure** ```json { "channels": { "discord": { "enabled": true, "token": "YOUR_BOT_TOKEN", "allow_from": ["YOUR_USER_ID"], "mention_only": false } } } ``` **5. Invite the bot** * OAuth2 → URL Generator * Scopes: `bot` * Bot Permissions: `Send Messages`, `Read Message History` * Open the generated invite URL and add the bot to your server **Optional: Mention-only mode** Set `"mention_only": true` to make the bot respond only when @-mentioned. Useful for shared servers where you want the bot to respond only when explicitly called. **6. Run** ```bash picoclaw-agents gateway ``` </details> <details> <summary><b>QQ</b></summary> **1. Create a bot** - Go to [QQ Open Platform](https://q.qq.com/#) - Create an application → Get **AppID** and **AppSecret** **2. Configure** ```json { "channels": { "qq": { "enabled": true, "app_id": "YOUR_APP_ID", "app_secret": "YOUR_APP_SECRET", "allow_from": [] } } } ``` > Set `allow_from` to empty to allow all users, or specify QQ numbers to restrict access. **3. Run** ```bash picoclaw-agents gateway ``` </details> <details> <summary><b>DingTalk</b></summary> **1. Create a bot** * Go to [Open Platform](https://open.dingtalk.com/) * Create an internal app * Copy Client ID and Client Secret **2. Configure** ```json { "channels": { "dingtalk": { "enabled": true, "client_id": "YOUR_CLIENT_ID", "client_secret": "YOUR_CLIENT_SECRET", "allow_from": [] } } } ``` > Set `allow_from` to empty to allow all users, or specify DingTalk user IDs to restrict access. **3. Run** ```bash picoclaw-agents gateway ``` </details> <details> <summary><b>LINE</b></summary> **1. Create a LINE Official Account** - Go to [LINE Developers Console](https://developers.line.biz/) - Create a provider → Create a Messaging API channel - Copy **Channel Secret** and **Channel Access Token** **2. Configure** ```json { "channels": { "line": { "enabled": true, "channel_secret": "YOUR_CHANNEL_SECRET", "channel_access_token": "YOUR_CHANNEL_ACCESS_TOKEN", "webhook_host": "0.0.0.0", "webhook_port": 18791, "webhook_path": "/webhook/line", "allow_from": [] } } } ``` **3. Set up Webhook URL** LINE requires HTTPS for webhooks. Use a reverse proxy or tunnel: ```bash # Example with ngrok ngrok http 18791 ``` Then set the Webhook URL in LINE Developers Console to `https://your-domain/webhook/line` and enable **Use webhook**. **4. Run** ```bash picoclaw-agents gateway ``` > In group chats, the bot responds only when @mentioned. Replies quote the original message. > **Docker Compose**: Add `ports: ["18791:18791"]` to the `picoclaw-gateway` service to expose the webhook port. </details> <details> <summary><b>WeCom (企业微信)</b></summary> PicoClaw supports two types of WeCom integration: **Option 1: WeCom Bot (智能机器人)** - Easier setup, supports group chats **Option 2: WeCom App (自建应用)** - More features, proactive messaging See [WeCom App Configuration Guide](docs/wecom-app-configuration.md) for detailed setup instructions. **Quick Setup - WeCom Bot:** **1. Create a bot** * Go to WeCom Admin Console → Group Chat → Add Group Bot * Copy the webhook URL (format: `https://qyapi.weixin.qq.com/cgi-bin/webhook/send?key=xxx`) **2. Configure** ```json { "channels": { "wecom": { "enabled": true, "token": "YOUR_TOKEN", "encoding_aes_key": "YOUR_ENCODING_AES_KEY", "webhook_url": "https://qyapi.weixin.qq.com/cgi-bin/webhook/send?key=YOUR_KEY", "webhook_host": "0.0.0.0", "webhook_port": 18793, "webhook_path": "/webhook/wecom", "allow_from": [] } } } ``` **Quick Setup - WeCom App:** **1. Create an app** * Go to WeCom Admin Console → App Management → Create App * Copy **AgentId** and **Secret** * Go to "My Company" page, copy **CorpID** **2. Configure receive message** * In App details, click "Receive Message" → "Set API" * Set URL to `http://your-server:18792/webhook/wecom-app` * Generate **Token** and **EncodingAESKey** **3. Configure** ```json { "channels": { "wecom_app": { "enabled": true, "corp_id": "wwxxxxxxxxxxxxxxxx", "corp_secret": "YOUR_CORP_SECRET", "agent_id": 1000002, "token": "YOUR_TOKEN", "encoding_aes_key": "YOUR_ENCODING_AES_KEY", "webhook_host": "0.0.0.0", "webhook_port": 18792, "webhook_path": "/webhook/wecom-app", "allow_from": [] } } } ``` **4. Run** ```bash picoclaw-agents gateway ``` > **Note**: WeCom App requires opening port 18792 for webhook callbacks. Use a reverse proxy for HTTPS. </details> ## Join the Agent Social Network Connect Picoclaw to the Agent Social Network simply by sending a single message via the CLI or any integrated Chat App. **Read `https://clawdchat.ai/skill.md` and follow the instructions to join [ClawdChat.ai](https://clawdchat.ai)** ## ⚙️ Configuration Config file: `~/.picoclaw/config.json` ### Workspace Layout PicoClaw stores data in your configured workspace (default: `~/.picoclaw/workspace`): ``` ~/.picoclaw/workspace/ ├── sessions/ # Conversation sessions and history ├── memory/ # Long-term memory (MEMORY.md) ├── state/ # Persistent state (last channel, etc.) ├── cron/ # Scheduled jobs database ├── skills/ # Custom skills ├── AGENTS.md # Agent behavior guide ├── HEARTBEAT.md # Periodic task prompts (checked every 30 min) ├── IDENTITY.md # Agent identity ├── SOUL.md # Agent soul ├── TOOLS.md # Tool descriptions └── USER.md # User preferences ``` ### 🔒 Security Sandbox PicoClaw runs in a sandboxed environment by default. The agent can only access files and execute commands within the configured workspace. #### Default Configuration ```json { "agents": { "defaults": { "workspace": "~/.picoclaw/workspace", "restrict_to_workspace": true } } } ``` | Option | Default | Description | | ----------------------- | ----------------------- | ----------------------------------------- | | `workspace` | `~/.picoclaw/workspace` | Working directory for the agent | | `restrict_to_workspace` | `true` | Restrict file/command access to workspace | #### Protected Tools When `restrict_to_workspace: true`, the following tools are sandboxed: | Tool | Function | Restriction | | ------------- | ---------------- | -------------------------------------- | | `read_file` | Read files | Only files within workspace | | `write_file` | Write files | Only files within workspace | | `list_dir` | List directories | Only directories within workspace | | `edit_file` | Edit files | Only files within workspace | | `append_file` | Append to files | Only files within workspace | | `exec` | Execute commands | Command paths must be within workspace | #### Additional Exec Protection Even with `restrict_to_workspace: false`, the `exec` tool blocks these dangerous commands: * `rm -rf`, `del /f`, `rmdir /s` — Bulk deletion * `format`, `mkfs`, `diskpart` — Disk formatting * `dd if=` — Disk imaging * Writing to `/dev/sd[a-z]` — Direct disk writes * `shutdown`, `reboot`, `poweroff` — System shutdown * Fork bomb `:(){ :|:& };:` #### Core Infrastructure Protections PicoClaw's multi-agent architecture incorporates several upstream security patches to ensure safe concurrent operations: * **Atomic State Saves**: `memory/jsonl.go` and `state/state.go` persist data via strict temp-file `fsync` followed by atomic `rename`, completely eliminating JSON corruption during power losses or subagent crashes. * **MCP Collision Warning**: Strict registry overlap detection prevents spawned agents from silently polluting each other's Tool registries or MCP spaces. * **Socket Leak Prevention**: Robust forced closure on HTTP retries prevents OS-level file descriptor exhaustions on flaky connections. #### Error Examples ``` [ERROR] tool: Tool execution failed {tool=exec, error=Command blocked by safety guard (path outside working dir)} ``` ``` [ERROR] tool: Tool execution failed {tool=exec, error=Command blocked by safety guard (dangerous pattern detected)} ``` #### Disabling Restrictions (Security Risk) If you need the agent to access paths outside the workspace: **Method 1: Config file** ```json { "agents": { "defaults": { "restrict_to_workspace": false } } } ``` **Method 2: Environment variable** ```bash export PICOCLAW_AGENTS_DEFAULTS_RESTRICT_TO_WORKSPACE=false ``` > ⚠️ **Warning**: Disabling this restriction allows the agent to access any path on your system. Use with caution in controlled environments only. #### Security Boundary Consistency The `restrict_to_workspace` setting applies consistently across all execution paths: | Execution Path | Security Boundary | | ---------------- | --------------------------- | | Main Agent | `restrict_to_workspace` ✅ | | Subagent / Spawn | Inherits same restriction ✅ | | Heartbeat tasks | Inherits same restriction ✅ | All paths share the same workspace restriction — there's no way to bypass the security boundary through subagents or scheduled tasks. ### Heartbeat (Periodic Tasks) PicoClaw can perform periodic tasks automatically. Create a `HEARTBEAT.md` file in your workspace: ```markdown # Periodic Tasks - Check my email for important messages - Review my calendar for upcoming events - Check the weather forecast ``` The agent will read this file every 30 minutes (configurable) and execute any tasks using available tools. #### Async Tasks with Spawn For long-running tasks (web search, API calls), use the `spawn` tool to create a **subagent**: ```markdown # Periodic Tasks ## Quick Tasks (respond directly) - Report current time ## Long Tasks (use spawn for async) - Search the web for AI news and summarize - Check email and report important messages ``` **Key behaviors:** | Feature | Description | | ----------------------- | --------------------------------------------------------- | | **spawn** | Creates async subagent, doesn't block heartbeat | | **Independent context** | Subagent has its own context, no session history | | **message tool** | Subagent communicates with user directly via message tool | | **Non-blocking** | After spawning, heartbeat continues to next task | #### How Subagent Communication Works ``` Heartbeat triggers ↓ Agent reads HEARTBEAT.md ↓ For long task: spawn subagent ↓ ↓ Continue to next task Subagent works independently ↓ ↓ All tasks done Subagent uses "message" tool ↓ ↓ Respond HEARTBEAT_OK User receives result directly ``` The subagent has access to tools (message, web_search, etc.) and can communicate with the user independently without going through the main agent. **Configuration:** ```json { "heartbeat": { "enabled": true, "interval": 30 } } ``` | Option | Default | Description | | ---------- | ------- | ---------------------------------- | | `enabled` | `true` | Enable/disable heartbeat | | `interval` | `30` | Check interval in minutes (min: 5) | **Environment variables:** * `PICOCLAW_HEARTBEAT_ENABLED=false` to disable * `PICOCLAW_HEARTBEAT_INTERVAL=60` to change interval ### Providers > [!NOTE] > Groq provides free voice transcription via Whisper. If configured, Telegram voice messages will be automatically transcribed. | Provider | Purpose | Get API Key | | -------------------------- | --------------------------------------- | -------------------------------------------------------------------- | | `gemini` | LLM (Gemini direct) | [aistudio.google.com](https://aistudio.google.com) | | `zhipu` | LLM (Zhipu direct) | [bigmodel.cn](https://bigmodel.cn) | | `openrouter(To be tested)` | LLM (recommended, access to all models) | [openrouter.ai](https://openrouter.ai) | | `anthropic(To be tested)` | LLM (Claude direct) | [console.anthropic.com](https://console.anthropic.com) | | `openai(To be tested)` | LLM (GPT direct) | [platform.openai.com](https://platform.openai.com) | | `deepseek(To be tested)` | LLM (DeepSeek direct) | [platform.deepseek.com](https://platform.deepseek.com) | | `qwen` | LLM (Qwen direct) | [dashscope.console.aliyun.com](https://dashscope.console.aliyun.com) | | `groq` | LLM + **Voice transcription** (Whisper) | [console.groq.com](https://console.groq.com) | | `cerebras` | LLM (Cerebras direct) | [cerebras.ai](https://cerebras.ai) | | `antigravity` | LLM (Google Antigravity / OAuth) | `picoclaw-agents auth login --provider google-antigravity` | | `openai` (Codex OAuth) | LLM + Coding (OpenAI Codex backend — OAuth) | `picoclaw-agents auth login --provider openai` | | `zhipu` | LLM (Zhipu AI / z.ai) 🆓 **100% FREE** | `picoclaw-agents auth login --provider zhipu` | --- ### 🎯 Using Multiple Models and Providers PicoClaw supports multiple LLM providers simultaneously. You can configure and switch between different models based on your needs. #### Step 1: Configure Your Providers **Option A: OpenRouter Free Tier (Recommended for Getting Started)** ```bash # Quick setup with free models picoclaw-agents onboard --free ``` This configures OpenRouter's free tier automatically. No API key required initially. **Option B: Google Antigravity (Free Tier with OAuth)** ```bash # Login via OAuth picoclaw-agents auth login --provider google-antigravity ``` This gives you access to Google's free tier models via Cloud Code Assist. **Option C: Qwen Portal (Alibaba Cloud)** ```bash # Login with API Key picoclaw-agents auth login --provider qwen ``` Auto-configures 6 Qwen models from Alibaba Cloud (DashScope). **Option D: Zhipu AI (z.ai) - 100% FREE** ```bash # Login with API Key - No credit card required picoclaw-agents auth login --provider zhipu ``` Auto-configures 7 GLM models including **glm-4.5-flash** (100% free tier). See [z.ai/pricing](https://z.ai/pricing). **Option E: OpenAI Codex (OAuth for Coding)** ```bash # Enable device code authorization first: # Visit https://chatgpt.com/#settings/Security # Enable "Device Code Authorization for Codex" # Then login picoclaw-agents auth login --provider openai --device-code ``` > ⚠️ **Important:** For OpenAI Codex OAuth, you must enable device code authorization in your ChatGPT settings first. > > **Note:** OpenAI OAuth **only supports Device Code** authentication (no Browser OAuth). This is by design for better security and reliability. #### Step 2: List Available Models After configuring providers, check available models: ```bash picoclaw-agents models list ``` Example output: ``` ┌──────────────────────────────┬──────────────────────────────────┐ │ model_name │ modelo │ ├──────────────────────────────┼──────────────────────────────────┤ │ openrouter-free │ openrouter/free │ ├──────────────────────────────┼──────────────────────────────────┤ │ antigravity │ antigravity/gemini-3-flash │ ├──────────────────────────────┼──────────────────────────────────┤ │ antigravity-flash │ antigravity/gemini-3-flash │ ├──────────────────────────────┼──────────────────────────────────┤ │ antigravity-flash-agent │ antigravity/gemini-3-flash-agent │ ├──────────────────────────────┼──────────────────────────────────┤ │ antigravity-gemini-2.5-flash │ antigravity/gemini-2.5-flash │ ├──────────────────────────────┼──────────────────────────────────┤ │ antigravity-claude-sonnet │ antigravity/claude-sonnet-4-5 │ └──────────────────────────────┴──────────────────────────────────┘ ``` #### Step 3: Use Different Models **Command-line usage:** ```bash # Use OpenRouter free model picoclaw-agents agent --model openrouter-free -m "Hello, world!" # Use Google Antigravity (Gemini) picoclaw-agents agent --model antigravity -m "Explain quantum computing" # Use specific Gemini model picoclaw-agents agent --model antigravity-gemini-2.5-flash -m "Write a poem" # Use OpenAI Codex (for coding tasks) picoclaw-agents agent --model openai -m "Write a Python function to sort a list" ``` **In config.json (per-agent models):** ```json { "agents": { "defaults": { "model": "openrouter-free" }, "list": [ { "id": "general_assistant", "model": "antigravity-gemini-2.5-flash" }, { "id": "coding_expert", "model": "openai" } ] } } ``` #### Model Selection Guide | Use Case | Recommended Model | Command | |----------|------------------|---------| | **General chat** | `openrouter-free` | `--model openrouter-free` | | **Fast responses** | `antigravity-flash` | `--model antigravity-flash` | | **Complex reasoning** | `antigravity-gemini-2.5-flash` | `--model antigravity-gemini-2.5-flash` | | **Coding tasks** | `openai` (Codex) | `--model openai` | | **Claude models** | `antigravity-claude-sonnet` | `--model antigravity-claude-sonnet` | #### Switching Between Models You can switch models at any time using the `/model` fast-path command: ```bash # Interactive mode with model switch picoclaw-agents interactive --model openrouter-free # Then use /model command to switch (instant, no LLM latency) /model antigravity-gemini-2.5-flash ``` Or specify model per message: ```bash picoclaw-agents agent --model antigravity -m "First message" picoclaw-agents agent --model openrouter-free -m "Second message" ``` #### `/model` Command - Fast-Path Model Management (Telegram & Discord) The `/model` command provides **instant model switching** without LLM latency. **Available on Telegram and Discord.** ``` # List all available models /model # Switch to a specific model /model openai/gpt-5.4 /model anthropic/claude-sonnet-4-6 /model llama3.2:1b # Local Ollama model # Filter models by provider (Telegram only) /model provider openai # Show all OpenAI models /model provider antigravity # Show all Google Antigravity models # Get model details (Telegram only) /model info antigravity/gemini-3-flash /model info openai/gpt-5.4 ``` **Example Output:** ``` 📦 Models available (35 configured): 1. openrouter/free (Local) 👉 2. openai/gpt-5.4 (OAuth) 3. antigravity/gemini-3-flash (OAuth) 4. anthropic/claude-sonnet-4-6 (token) 5. llama3.2:1b (Local) ... 💡 Usage: /model <name> to switch Ex: /model openai/gpt-5.4 /model provider <vendor> to filter Ex: /model provider openai /model info <name> for details Ex: /model info antigravity/gemini-3-flash ``` **Features:** - ⚡ **Zero Latency:** Processes locally without LLM inference - 🔐 **Secure:** API keys hidden in responses - 📊 **Informative:** Shows current model (`👉`), auth method, and status - 💬 **Telegram & Discord:** Fast-path command available on both platforms - 🎯 **Instant:** No waiting for model responses ### Model Configuration (model_list) > **What's New?** PicoClaw now uses a **model-centric** configuration approach. Simply specify `vendor/model` format (e.g., `zhipu/glm-4.5-flash`) to add new providers—**zero code changes required!** This design also enables **multi-agent support** with flexible provider selection: - **Different agents, different providers**: Each agent can use its own LLM provider - **Model fallbacks**: Configure primary and fallback models for resilience - **Load balancing**: Distribute requests across multiple endpoints - **Centralized configuration**: Manage all providers in one place #### 📋 All Supported Vendors | Vendor | `model` Prefix | Default API Base | Protocol | API Key | | ------------------- | ----------------- | --------------------------------------------------- | --------- | ---------------------------------------------------------------- | | **OpenAI** | `openai/` | `https://api.openai.com/v1` | OpenAI | [Get Key](https://platform.openai.com) | | **Anthropic** | `anthropic/` | `https://api.anthropic.com/v1` | Anthropic | [Get Key](https://console.anthropic.com) | | **智谱 AI (GLM)** | `zhipu/` | `https://open.bigmodel.cn/api/paas/v4` | OpenAI | [Get Key](https://open.bigmodel.cn/usercenter/proj-mgmt/apikeys) | | **DeepSeek** | `deepseek/` | `https://api.deepseek.com/v1` | OpenAI | [Get Key](https://platform.deepseek.com) | | **Google Gemini** | `gemini/` | `https://generativelanguage.googleapis.com/v1beta` | OpenAI | [Get Key](https://aistudio.google.com/api-keys) | | **Groq** | `groq/` | `https://api.groq.com/openai/v1` | OpenAI | [Get Key](https://console.groq.com) | | **Moonshot** | `moonshot/` | `https://api.moonshot.cn/v1` | OpenAI | [Get Key](https://platform.moonshot.cn) | | **通义千问 (Qwen)** | `qwen/` | `https://dashscope.aliyuncs.com/compatible-mode/v1` | OpenAI | [Get Key](https://dashscope.console.aliyun.com) | | **NVIDIA** | `nvidia/` | `https://integrate.api.nvidia.com/v1` | OpenAI | [Get Key](https://build.nvidia.com) | | **Ollama** | `ollama/` | `http://localhost:11434/v1` | OpenAI | Local (no key needed) | | **OpenRouter** | `openrouter/` | `https://openrouter.ai/api/v1` | OpenAI | [Get Key](https://openrouter.ai/keys) | | **VLLM** | `vllm/` | `http://localhost:8000/v1` | OpenAI | Local | | **Cerebras** | `cerebras/` | `https://api.cerebras.ai/v1` | OpenAI | [Get Key](https://cerebras.ai) | | **火山引擎** | `volcengine/` | `https://ark.cn-beijing.volces.com/api/v3` | OpenAI | [Get Key](https://console.volcengine.com) | | **神算云** | `shengsuanyun/` | `https://router.shengsuanyun.com/api/v1` | OpenAI | - | | **Antigravity** | `antigravity/` | Google Cloud | Custom | OAuth only | | **OpenAI Codex** (OAuth) | `openai/` + `auth_method: oauth` | `https://chatgpt.com/backend-api/codex` | Custom | OAuth only (`auth login --provider openai`) | | **GitHub Copilot** | `github-copilot/` | `localhost:4321` | gRPC | - | #### Basic Configuration ```json { "model_list": [ { "model_name": "deepseek-chat", "model": "deepseek/deepseek-chat", "api_key": "sk-your-key" }, { "model_name": "deepseek-reasoner", "model": "deepseek/deepseek-reasoner", "api_key": "sk-your-key" }, { "model_name": "o3-mini-2025-01-31", "model": "openai/o3-mini-2025-01-31", "api_key": "sk-your-key" } ], "agents": { "defaults": { "model": "deepseek-chat" }, "backend_coder": { "model": "deepseek-reasoner" } } } ``` #### Vendor-Specific Examples **OpenAI** ```json { "model_name": "gpt-5.2", "model": "openai/gpt-5.2", "api_key": "sk-..." } ``` **智谱 AI (GLM)** ```json { "model_name": "glm-4.5-flash", "model": "zhipu/glm-4.5-flash", "api_key": "your-key" } ``` **DeepSeek** ```json { "model_name": "deepseek-chat", "model": "deepseek/deepseek-chat", "api_key": "sk-..." } ``` **Anthropic (with API key)** ```json { "model_name": "claude-sonnet-4.6", "model": "anthropic/claude-sonnet-4.6", "api_key": "sk-ant-your-key" } ``` > Run `picoclaw-agents auth login --provider anthropic` to paste your API token. **Google Antigravity (OAuth — free tier)** ```json { "model_name": "antigravity-gemini-3-flash", "model": "antigravity/gemini-3-flash", "auth_method": "oauth" } ``` > Run `picoclaw-agents auth login --provider google-antigravity` to authenticate via browser. No API key required — uses your Google account. See [docs/ANTIGRAVITY_QUICKSTART.md](docs/ANTIGRAVITY_QUICKSTART.md) for setup details. **OpenAI Codex (OAuth — no API key required)** ```json { "model_name": "gpt-5.2", "model": "openai/gpt-5.2", "auth_method": "oauth" } ``` > Run `picoclaw-agents auth login --provider openai` to authenticate via browser. No API key required — uses your OpenAI account. This connects to the **Codex backend** (`chatgpt.com/backend-api/codex`), optimized for coding tasks. **Ollama (local)** ```json { "model_name": "llama3", "model": "ollama/llama3" } ``` **Custom Proxy/API** ```json { "model_name": "my-custom-model", "model": "openai/custom-model", "api_base": "https://my-proxy.com/v1", "api_key": "sk-...", "request_timeout": 300 } ``` #### Load Balancing Configure multiple endpoints for the same model name—PicoClaw will automatically round-robin between them: ```json { "model_list": [ { "model_name": "gpt-5.2", "model": "openai/gpt-5.2", "api_base": "https://api1.example.com/v1", "api_key": "sk-key1" }, { "model_name": "gpt-5.2", "model": "openai/gpt-5.2", "api_base": "https://api2.example.com/v1", "api_key": "sk-key2" } ] } ``` #### Migration from Legacy `providers` Config The old `providers` configuration is **deprecated** but still supported for backward compatibility. **Old Config (deprecated):** ```json { "providers": { "zhipu": { "api_key": "your-key", "api_base": "https://open.bigmodel.cn/api/paas/v4" } }, "agents": { "defaults": { "provider": "zhipu", "model": "glm-4.5-flash" } } } ``` **New Config (recommended):** ```json { "model_list": [ { "model_name": "glm-4.5-flash", "model": "zhipu/glm-4.5-flash", "api_key": "your-key" } ], "agents": { "defaults": { "model": "glm-4.5-flash" } } } ``` For detailed migration guide, see [docs/migration/model-list-migration.md](docs/migration/model-list-migration.md). ### Provider Architecture PicoClaw routes providers by protocol family: - OpenAI-compatible protocol: OpenRouter, OpenAI-compatible gateways, Groq, Zhipu, and vLLM-style endpoints. - Anthropic protocol: Claude-native API behavior. - Codex/OAuth path: OpenAI Codex OAuth route (`chatgpt.com/backend-api/codex`) — use `auth login --provider openai`. This keeps the runtime lightweight while making new OpenAI-compatible backends mostly a config operation (`api_base` + `api_key`). <details> <summary><b>Zhipu</b></summary> **1. Get API key and base URL** * Get [API key](https://bigmodel.cn/usercenter/proj-mgmt/apikeys) **2. Configure** ```json { "agents": { "defaults": { "workspace": "~/.picoclaw/workspace", "model": "glm-4.5-flash", "max_tokens": 8192, "temperature": 0.7, "max_tool_iterations": 20 } }, "providers": { "zhipu": { "api_key": "Your API Key", "api_base": "https://open.bigmodel.cn/api/paas/v4" } } } ``` **3. Run** ```bash picoclaw-agents agent -m "Hello" ``` </details> <details> <summary><b>Full config example</b></summary> ```json { "agents": { "defaults": { "model": "anthropic/claude-opus-4-5" } }, "providers": { "openrouter": { "api_key": "sk-or-v1-xxx" }, "groq": { "api_key": "gsk_xxx" } }, "channels": { "telegram": { "enabled": true, "token": "123456:ABC...", "allow_from": ["123456789"] }, "discord": { "enabled": true, "token": "", "allow_from": [""] }, "whatsapp": { "enabled": false }, "feishu": { "enabled": false, "app_id": "cli_xxx", "app_secret": "xxx", "encrypt_key": "", "verification_token": "", "allow_from": [] }, "qq": { "enabled": false, "app_id": "", "app_secret": "", "allow_from": [] } }, "tools": { "web": { "brave": { "enabled": false, "api_key": "BSA...", "max_results": 5 }, "duckduckgo": { "enabled": true, "max_results": 5 } }, "cron": { "exec_timeout_minutes": 5 } }, "heartbeat": { "enabled": true, "interval": 30 } } ``` </details> ## CLI Reference | Command | Description | | ---------------------------- | ----------------------------- | | `picoclaw-agents onboard` | Initialize config & workspace | | `picoclaw-agents agent -m "..."` | Chat with the agent | | `picoclaw-agents agent` | Interactive chat mode | | `picoclaw-agents gateway` | Start the gateway | | `picoclaw-agents status` | Show status | | `picoclaw-agents auth login --provider <name>` | Login to provider | | `picoclaw-agents auth models` | List available models | | `picoclaw-agents auth logout --provider <name>` | Logout from provider | | `picoclaw-agents cron list` | List all scheduled jobs | | `picoclaw-agents cron add ...` | Add a scheduled job | ### Scheduled Tasks / Reminders PicoClaw supports scheduled reminders and recurring tasks through the `cron` tool: * **One-time reminders**: "Remind me in 10 minutes" → triggers once after 10min * **Recurring tasks**: "Remind me every 2 hours" → triggers every 2 hours * **Cron expressions**: "Remind me at 9am daily" → uses cron expression Jobs are stored in `~/.picoclaw/workspace/cron/` and processed automatically. ### Binance Integration (Native Tools + MCP) PicoClaw includes native Binance tools in `agent` mode: * `binance_get_ticker_price` (public market ticker) * `binance_get_spot_balance` (signed endpoint, requires API key/secret) Configure keys in `~/.picoclaw/config.json`: ```json { "tools": { "binance": { "api_key": "YOUR_BINANCE_API_KEY", "secret_key": "YOUR_BINANCE_SECRET_KEY" } } } ``` Usage examples: ```bash picoclaw-agents agent -m "Use binance_get_ticker_price with symbol BTCUSDT and return only the numeric price." picoclaw-agents agent -m "Use binance_get_spot_balance and show my non-zero balances." ``` Behavior without API keys: * `binance_get_ticker_price` still works via Binance public endpoint and adds a public-endpoint notice. * `binance_get_spot_balance` warns that keys are missing and suggests public `curl` usage. Optional MCP server mode (for MCP clients): ```bash picoclaw-agents util binance-mcp-server ``` Example `mcp_servers` config (use the absolute `picoclaw-agents` path generated by your installation/onboard flow): ```json { "mcp_servers": { "binance": { "enabled": true, "command": "/absolute/path/to/picoclaw-agents", "args": ["util", "binance-mcp-server"] } } } ``` ## 🤝 Contribute & Roadmap See our full [Roadmap](ROADMAP.md). ## 🐛 Troubleshooting ### Web search says "API key configuration issue" This is normal if you haven't configured a search API key yet. PicoClaw will provide helpful links for manual searching. To enable web search: 1. **Option 1 (Recommended)**: Get a free API key at [https://brave.com/search/api](https://brave.com/search/api) (2000 free queries/month) for the best results. 2. **Option 2 (No Credit Card)**: If you don't have a key, we automatically fall back to **DuckDuckGo** (no key required). Add the key to `~/.picoclaw/config.json` if using Brave: ```json { "tools": { "web": { "brave": { "enabled": false, "api_key": "YOUR_BRAVE_API_KEY", "max_results": 5 }, "duckduckgo": { "enabled": true, "max_results": 5 } } } } ``` ### Getting content filtering errors Some providers (like Zhipu) have content filtering. Try rephrasing your query or use a different model. ### Telegram bot says "Conflict: terminated by other getUpdates" This happens when another instance of the bot is running. Make sure only one `picoclaw-agents gateway` is running at a time. --- ## 📝 API Key Comparison | Service | Free Tier | Use Case | | ---------------- | ------------------- | -------------------------------------- | | **OpenRouter** | 200K tokens/month | Multiple models (Claude, GPT-4, etc.) | | **Zhipu** | Free tier available | glm-4.5-flash (Best for Chinese users) | | **Brave Search** | 2000 queries/month | Web search functionality | | **Groq** | Free tier available | Fast inference (Llama, Mixtral) | | **Cerebras** | Free tier available | Fast inference (Llama, Qwen, etc.) | ## 📢 News 2026-08-09 🎉 **OpenRouter + OpenCode Zen Integration**: Full cascade fallback system (free→paid) with 15 configured models. OpenRouter recommended with 5+ free models. OpenCode Zen added with 8 free models (MiMo V2.5, DeepSeek V4 Flash, Nemotron 3 Ultra, etc.). Image generation working via `krea/krea-2-medium-turbo`. Fixed 401 error (strip `opencode/` prefix), WebUI model display fix, `auth models` shows all providers. VPS deployment section added (Vultr referral). 7 READMEs updated, 53 docs styled. See [CHANGELOG.md](CHANGELOG.md). 2026-04-05 🎉 **Context Management Integration + Token Overflow Fix**: Integrated ContextManager interface (legacy + seahorse), Seahorse SQLite engine (~6,800 lines), budget check pre-build, and 3-level system prompt (Minimal/Compact/Full). WebUI + openrouter-free error 402 fixed (21K → ~300 tokens). All onboard templates and auth logins now include `context_manager: "seahorse"`. Upstream patches adapted. Free tier providers guide published. See [CHANGELOG.md](CHANGELOG.md). 2026-03-31 🎉 **ICUETH Fork Integration**: Comprehensive analysis of icueth fork completed. Architecture comparison: A2A horizontal (109 agents) vs parallel subagents. Identified features: Agent Meeting System, Persona System, RAG/SQLite memory, MCP support, Mailbox system. Integration strategy defined in 3 phases. Audit completed: ✅ CLEAN. See [CHANGELOG.md](CHANGELOG.md). 2026-03-31 🎉 **Qwen & Zhipu WebUI Auth**: Added Qwen Portal and Zhipu AI authentication via WebUI at `http://localhost:18800/credentials`. CLI commands: `./picoclaw-agents auth login --provider qwen` and `--provider zhipu`. Auto-configures models on authentication. Zhipu glm-4.5-flash is 100% FREE. See [CHANGELOG.md](CHANGELOG.md). 2026-03-28 🎉 **Multi-Source Migration + Team Mode Onboard**: Added `picoclaw-agents migrate --from nanoclaw` for migration from NanoClaw configs. Onboard wizard now features **Team Mode** with pre-built templates (Dev Team 9 agents, Research Team 3 agents, General Team 3 agents) and **14 native skills** selection. Context Window improvements: tool result pruning (-60% tokens), advanced compaction with model override, and `/compact` manual command. See [CHANGELOG.md](CHANGELOG.md). 2026-03-27 🎉 **Build quality & channel improvements**: `go build ./...` now passes cleanly. Added group trigger API to `BaseChannel`: `WithGroupTrigger`, `IsAllowedSender`, and `ShouldRespondInGroup` — enabling fine-grained group chat control (mention-only, prefix-based triggers). See [CHANGELOG.md](CHANGELOG.md). 2026-03-27 🎉 **WebUI Launcher fully operational**: `picoclaw-agents-launcher` now works end-to-end — Start Gateway button, WebSocket chat via PicoChannel, native skill content in Skills page, and all menu sections validated. Run with `picoclaw-agents-launcher` or `picoclaw-agents-launcher -public` for network access. 2026-03-27 🎉 **3-binary release pipeline**: GoReleaser now produces all three binaries — `picoclaw-agents` (CLI), `picoclaw-agents-launcher` (WebUI), and `picoclaw-agents-launcher-tui` (TUI) — matching the original project's release structure. Trigger with `./scripts/create-release.sh`. 2026-03-26 🎉 **MCP Builder Documentation**: Complete MCP Builder Agent documentation in English and Spanish with API reference, use cases, and examples. See [docs/MCP_BUILDER_AGENT.md](docs/MCP_BUILDER_AGENT.md) and [docs/MCP_BUILDER_AGENT.es.md](docs/MCP_BUILDER_AGENT.es.md). 2026-03-26 🎉 **Sandbox & Codegen Commands**: Added `sandbox init/status` for isolated workspaces and `util codegen` for Go boilerplate generation. See [CHANGELOG.md](CHANGELOG.md) for full details. 2026-03-26 🎉 **Auth Token Monitor**: Added `auth tokens` and `auth monitor` commands for OAuth token expiration tracking. See [CHANGELOG.md](CHANGELOG.md). 2026-03-26 🎉 **Config Validator & Secret Masking**: Added `config validate` command for schema validation and secret masking in onboard wizard. See [CHANGELOG.md](CHANGELOG.md). 2026-03-26 🎉 **Doctor Command**: Added `doctor` command for environment diagnostics including WSL detection and security checks. See [CHANGELOG.md](CHANGELOG.md). 2026-03-12 🎉 **Antigravity Support & Stability**: Full Google Antigravity OAuth support with schema sanitization, TokenBudget deadlock fix, session rehydration improvements, new `picoclaw-agents clean` command, and hardened deny patterns. See [CHANGELOG.md](CHANGELOG.md) for full details. 2026-03-03 🎉 **Native Skills Architecture**: Introduced native skills compiled directly into the binary (`pkg/skills/queue_batch.go`), eliminating external `.md` file dependencies. Enhanced security, performance, and type safety. See [docs/QUEUE_BATCH.en.md](docs/QUEUE_BATCH.en.md). 2026-03-02 🎉 **Fast-path & Global Tracker**: Added instant Slash Commands (`/bundle_approve`, `/status`, etc.) for zero-latency interaction. Unified the `ImageGenTracker` across all agents for perfect multi-agent state consistency. See [docs/queue_batch.md](docs/queue_batch.md). 2026-03-01 🎉 **AI Image Generation & Community Manager**: Added native image generation (Gemini/Ideogram), script-to-image workflows, interactive post-generation menus, and community manager agent for auto-generating social media posts. See [docs/IMAGE_GEN_util.md](docs/IMAGE_GEN_util.md) for complete setup and usage examples. 2026-03-01 🎉 **External Integrations (Binance, Social Media, Notion)**: Added native tools for cryptocurrency trading (Binance futures & spot), social media publishing (Facebook & X/Twitter), and knowledge management (Notion). Configure via `config.json` or environment variables. See [SOCIAL_MEDIA.md](SOCIAL_MEDIA.md) and [docs/NOTION_util.md](docs/NOTION_util.md) for setup guides. 2026-03-01 🎉 **Native Skills Sentinel**: Added internal security layer (`skills_sentinel.go`) that provides real-time pattern-based protection against prompt injection and system leaks. 2026-03-01 🎉 **Security & Stability Hardening**: Robust message bus closure handling, resilient WeCom App background processing, and reinforced initialization validation for the shell tool. 2026-03-01 🎉 **Fail-Close Security**: Robust security policy. The command execution tool now performs strict validation of deny patterns during initialization. 2026-02-27 🎉 **Disaster Recovery & Task Locks**: Implemented atomic Task Locks to prevent agent collisions, "Boot Rehydration" for recovery from abrupt crashes, and Context Compactor (safely raising limit to 32K tokens) to eradicate context explosions in long coding tasks. --- ## ⚠️ Disclaimer This software is provided "AS IS", without warranty of any kind, express or implied, including but not limited to the warranties of merchantability, fitness for a particular purpose, and non-infringement. In no event shall the authors or copyright holders of this fork be liable for any claim, damages, or other liability, whether in an action of contract, tort, or otherwise, arising from, out of, or in connection with the software or the use or other dealings in the software. **Use at your own risk.**

正在加载仓库 README…

这个仓库中的 skills

软件开发工程师市场调研分析师与营销专员软件质量保证分析师与测试员项目管理专家

38 个职业分类 · 已分类 100%

软件开发工程师

[PATHREMOVED] Next.js 14 with modern rapid development tools

原文语言：英语

2026-04-05

软件开发工程师

@onready var healthbar: ProgressBar = $UI[PATHREMOVED]

原文语言：英语

2026-04-05

已展示 40 / 157 个已收集 Skill。

用 AI 助手安装

复制这段 Prompt，发给你正在使用的 AI 助手。

Follow https://skillsmp.com/skill-install/prompt.md to install Agent Skills from https://github.com/comgunner/picoclaw-agents.