[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"project-807":3},{"id":4,"name":5,"fullName":6,"owner":5,"repo":5,"description":7,"homepage":8,"htmlUrl":9,"language":10,"languages":9,"totalLinesOfCode":9,"stars":11,"forks":12,"watchers":13,"openIssues":14,"contributorsCount":15,"subscribersCount":15,"size":15,"stars1d":16,"stars7d":17,"stars30d":18,"stars90d":15,"forks30d":15,"starsTrendScore":19,"compositeScore":20,"rankGlobal":9,"rankLanguage":9,"license":21,"archived":22,"fork":22,"defaultBranch":23,"hasWiki":24,"hasPages":22,"topics":25,"createdAt":9,"pushedAt":9,"updatedAt":39,"readmeContent":40,"aiSummary":41,"trendingCount":15,"starSnapshotCount":15,"syncStatus":42,"lastSyncTime":43,"discoverSource":44},807,"thClaws","thClaws\u002FthClaws","Open-source AI agent harness in native Rust — GUI, CLI, headless, and webapp from one binary. Multi-provider, MCP, skills, plugins, agent teams.","https:\u002F\u002Fthclaws.ai",null,"Rust",1085,153,17,1,0,4,27,214,12,19.56,"Apache License 2.0",false,"main",true,[26,27,28,29,30,31,32,33,34,35,36,37,38],"agent-harness","agent-teams","ai","ai-agent","ai-agents","ai-assistant","ai-coding","ai-swarm","claude-code","claude-cowork","multi-provider","rust","tauri","2026-06-12 02:00:18","# thClaws 🦞\n\n> **Open-source Agent Harness Platform** — a native AI agent workspace that codes, automates, remembers, and coordinates. Runs on your own machine. Sovereign by design.\n\nthClaws is a **native-Rust AI agent workspace** that runs locally on your machine. Not just coding — it edits code, automates workflows, searches your knowledge bases, and coordinates teams of agents, all in one binary. You tell it what you want in natural language; it reads your files, runs commands, uses tools, and talks back to you while it works.\n\n[![License: MIT OR Apache-2.0](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002Flicense-MIT%20OR%20Apache--2.0-blue.svg)](#license)\n[![Platform: macOS · Windows · Linux](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002Fplatform-macOS%20·%20Windows%20·%20Linux-lightgrey.svg)](#installation)\n[![Built with Rust](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002Fbuilt%20with-Rust-orange.svg)](https:\u002F\u002Fwww.rust-lang.org\u002F)\n\n---\n\n## See it work\n\nThree tabs, one binary — captured from a live thClaws session looking at its own source.\n\n\u003Ctable>\n  \u003Ctr>\n    \u003Ctd width=\"33%\" align=\"center\">\n      \u003Ca href=\"docs\u002Fimg\u002Fscreen-files.webp\">\u003Cimg src=\"docs\u002Fimg\u002Fscreen-files.webp\" alt=\"thClaws Files tab — landing-page source open in the editor with file tree\" \u002F>\u003C\u002Fa>\u003Cbr\u002F>\n      \u003Cstrong>Files\u003C\u002Fstrong>\u003Cbr\u002F>\n      \u003Csub>preview · edit · browse — codemirror + tiptap\u003C\u002Fsub>\n    \u003C\u002Ftd>\n    \u003Ctd width=\"33%\" align=\"center\">\n      \u003Ca href=\"docs\u002Fimg\u002Fscreen-terminal.webp\">\u003Cimg src=\"docs\u002Fimg\u002Fscreen-terminal.webp\" alt=\"thClaws Terminal tab — REPL session with ASCII banner\" \u002F>\u003C\u002Fa>\u003Cbr\u002F>\n      \u003Cstrong>Terminal\u003C\u002Fstrong>\u003Cbr\u002F>\n      \u003Csub>raw REPL · slash commands · ANSI tool output\u003C\u002Fsub>\n    \u003C\u002Ftd>\n    \u003Ctd width=\"33%\" align=\"center\">\n      \u003Ca href=\"docs\u002Fimg\u002Fscreen-chat.webp\">\u003Cimg src=\"docs\u002Fimg\u002Fscreen-chat.webp\" alt=\"thClaws Chat tab — conversational interface answering a product question\" \u002F>\u003C\u002Fa>\u003Cbr\u002F>\n      \u003Cstrong>Chat\u003C\u002Fstrong>\u003Cbr\u002F>\n      \u003Csub>conversational · markdown render · tool indicators\u003C\u002Fsub>\n    \u003C\u002Ftd>\n  \u003C\u002Ftr>\n\u003C\u002Ftable>\n\n---\n\n## Three interfaces, one binary\n\n- **Desktop GUI** (`thclaws`) — a native window with Terminal, Chat, Files, and optional Team tabs.\n- **CLI REPL** (`thclaws --cli`) — an interactive terminal prompt for SSH, headless servers, or when you want zero GUI overhead.\n- **Non-interactive mode** (`thclaws -p \"prompt\"`) — runs a single turn and exits. Handy for scripts, CI pipelines, and shell one-liners.\n\n---\n\n## What makes it different\n\n- **Multi-provider.** Anthropic, OpenAI, Gemini, Alibaba DashScope, OpenRouter, Ollama (local and Anthropic-compatible), Agentic Press, plus a generic **OpenAI-compatible** slot (`oai\u002F*`) for LiteLLM \u002F Portkey \u002F Helicone \u002F vLLM \u002F internal proxies — auto-detected by model name prefix. Switch mid-session with `\u002Fmodel` or swap the whole provider with `\u002Fprovider`.\n\n- **Any knowledge worker, not just engineers.** Chat tab for researchers, PMs, ops, legal, marketing, finance — natural-language prompts, file access, knowledge-base lookup, drafting. Terminal tab for engineers who want the raw REPL. Same engine, same sessions, same config — different preferred surface.\n\n- **Open standards, not a walled garden.** Built on the conventions the agent-tooling industry is converging on, not bespoke formats you have to learn only for us. [Model Context Protocol](https:\u002F\u002Fmodelcontextprotocol.io\u002F) for tool servers. [`AGENTS.md`](https:\u002F\u002Fagents.md) for project instructions — the vendor-neutral standard adopted by Google, OpenAI, Factory, Sourcegraph, and Cursor. `SKILL.md` with YAML frontmatter for packaged workflows. Your configuration is portable between thClaws, other agents that speak the same standards, and whatever comes next.\n\n- **Skills.** Reusable expert workflows packaged as a directory with `SKILL.md` plus optional scripts. The agent picks the right skill automatically when a request matches the `whenToUse` trigger, or you can invoke one explicitly as `\u002F\u003Cskill-name>`. Install from a git URL or `.zip` archive with `\u002Fskill install`.\n\n- **MCP servers.** Plug in tools built by third parties — GitHub, filesystems, databases, browsers, Slack, and more. Both stdio and HTTP Streamable transports, with OAuth 2.1 + PKCE for protected servers. Add one with `\u002Fmcp add` or ship a `.mcp.json` in your project.\n\n- **Plugin system.** Skills + commands + agent definitions + MCP servers bundled under a single manifest, installable from git or `.zip`. One install, one uninstall, one version to pin — ideal for sharing a team's extensions.\n\n- **Memory & project instructions.** Drop an `AGENTS.md` (or `CLAUDE.md`) in your repo — thClaws walks up from `cwd` and injects every match into the system prompt. A persistent memory store holds longer-lived facts the agent has learned about you, classified as `user` \u002F `feedback` \u002F `project` \u002F `reference` and stored as markdown you can read, edit, or commit.\n\n- **Knowledge bases (KMS).** Per-project and per-user wikis the agent can search and read on demand. Drop markdown pages under `.thclaws\u002Fkms\u002F\u003Cname>\u002Fpages\u002F`, give each a one-line entry in `index.md`, and the agent gets a table of contents every turn plus `KmsRead` \u002F `KmsSearch` tools. No embeddings — grep + read, following Andrej Karpathy's LLM-wiki pattern.\n\n- **Agent orchestration.** Delegate subtasks to isolated sub-agents via the `Task` tool — each gets its own tool registry and can recurse up to 3 levels deep. Scale further with **Agent Teams**: multiple thClaws processes coordinating through a shared mailbox and task queue, each in its own tmux pane and optional git worktree. One agent writes your backend while a teammate builds the frontend in parallel, lead merges the branches when both are done.\n\n- **Settings as one file.** Every knob — permission mode, thinking budget, allowed\u002Fdisallowed tools, provider endpoints, KMS attachments — lives in `.thclaws\u002Fsettings.json` (project) or `~\u002F.config\u002Fthclaws\u002Fsettings.json` (user). API keys go in the OS keychain by default (macOS Keychain \u002F Windows Credential Manager \u002F Linux Secret Service) with `.env` fallback for CI.\n\n- **Safety first.** A filesystem sandbox scopes file tools to the working directory. Destructive shell commands are flagged before execution. You approve every mutating tool call unless you've opted into auto-approve.\n\n- **Offline-capable.** Ollama (native and Anthropic-compatible) lets you run entirely against a local model — no cloud round-trip, no API key.\n\n- **Deploy what you build.** Ship the landing pages, web apps, APIs, and AI agents you create through [Agentic Press Hosting](https:\u002F\u002Fagentic-press.com) (partnered with SIS Cloud Service and Artech.Cloud) — or any other host you prefer. Schedule agents on cron, respond to webhooks, stream from public URLs. The deploy flow ships as a plugin (`\u002Fplugin install …-deploy`), so hosts are swappable; the client never locks you in.\n\n- **Shell escape.** Prefix any REPL line with `!` to run a shell command directly — no tokens, no approval prompt, no agent round-trip (`! git status`, `! ls`, etc.).\n\n---\n\n## Installation\n\n### Pre-built binaries\n\nDownload the latest release for your platform from the [Releases page](https:\u002F\u002Fgithub.com\u002FthClaws\u002FthClaws\u002Freleases) or from [thclaws.ai\u002Fdownloads](https:\u002F\u002Fthclaws.ai\u002Fdownloads.html).\n\nSupported: macOS (Apple Silicon & Intel), Windows (x86_64 & ARM64), Linux (x86_64 & ARM64).\n\n#### Linux runtime dependencies\n\nThe Linux GUI binary links against the Wayland and webkit2gtk client libraries at runtime. Most desktop distros (Ubuntu Desktop, Fedora Workstation, etc.) ship them by default. **Headless servers** (cloud VMs, AWS EC2, Docker images without a display) typically don't — `thclaws` will fail at startup with `error while loading shared libraries: libwayland-client.so.0`.\n\nTwo options on a headless box:\n\n**(a) Use CLI mode** — no GUI deps required:\n\n```sh\nthclaws --cli                       # interactive REPL\nthclaws -p \"what does src\u002Fmain.rs do?\"  # one-shot\n```\n\n**(b) Install the GUI deps** — only if you actually want to run the webview:\n\n```sh\n# Debian \u002F Ubuntu\nsudo apt install libwayland-client0 libwebkit2gtk-4.1-0 libsoup-3.0-0\n\n# Fedora \u002F RHEL\nsudo dnf install wayland libsoup3 webkit2gtk4.1\n```\n\n### Build from source\n\n**Prerequisites:** Rust 1.85+, Node.js 20+, pnpm 9+.\n\n```sh\ngit clone https:\u002F\u002Fgithub.com\u002FthClaws\u002FthClaws.git\ncd thClaws\n\n# Build frontend (React + Vite, bundled as a single HTML file)\ncd frontend && pnpm install && pnpm build && cd ..\n\n# Build Rust (CLI + GUI)\ncargo build --release --features gui --bin thclaws\n\n# Run\n.\u002Ftarget\u002Frelease\u002Fthclaws          # GUI\n.\u002Ftarget\u002Frelease\u002Fthclaws --cli    # CLI\n.\u002Ftarget\u002Frelease\u002Fthclaws -p \"what does src\u002Fmain.rs do?\"  # one-shot\n```\n\n---\n\n## Quick start\n\n```sh\n# First run: pick a secrets backend (OS keychain or .env) when prompted\nthclaws\n\n# Configure a provider (inside the REPL)\n❯ \u002Fprovider anthropic\n❯ \u002Fmodel claude-sonnet-4-6\n\n# Or try OpenRouter for 300+ models via one key\n❯ \u002Fprovider openrouter\n❯ \u002Fmodel openrouter\u002Fanthropic\u002Fclaude-sonnet-4-6\n\n# Drop an AGENTS.md or CLAUDE.md in your repo — it's read automatically\n\n# Useful slash commands\n❯ \u002Fhelp         # list everything\n❯ \u002Fmodels       # list available models for the current provider\n❯ \u002Fkms          # list attached knowledge bases\n❯ \u002Fskill install https:\u002F\u002Fgithub.com\u002Fanthropics\u002Fskills.git\n❯ \u002Fmcp add github https:\u002F\u002Fmcp.github.com\n❯ ! git status  # shell escape\n```\n\n---\n\n## Configuration\n\nthClaws reads settings in this precedence order (higher wins):\n\n1. CLI flags\n2. `.thclaws\u002Fsettings.json` (project)\n3. `~\u002F.config\u002Fthclaws\u002Fsettings.json` (user)\n4. `~\u002F.claude\u002Fsettings.json` (fallback location)\n5. Compiled-in defaults\n\nOpen-standard files are honored directly:\n\n- `CLAUDE.md` \u002F `AGENTS.md` — system prompt additions, walked up from `cwd`\n- `.thclaws\u002Fskills\u002F` \u002F `.claude\u002Fskills\u002F` — skill catalog\n- `.thclaws\u002Fagents\u002F` \u002F `.claude\u002Fagents\u002F` — subagent definitions\n- `.mcp.json` \u002F `.thclaws\u002Fmcp.json` — MCP server configuration\n- `.thclaws-plugin\u002Fplugin.json` \u002F `.claude-plugin\u002Fplugin.json` — plugin manifest\n\nAPI keys are **never stored in config files** — only in the OS keychain (default) or `.env`.\n\n---\n\n## Documentation\n\n- **Official site** — [thclaws.ai](https:\u002F\u002Fthclaws.ai)\n- **Full user manual** — [thclaws.ai\u002Fmanual](https:\u002F\u002Fthclaws.ai\u002Fmanual) *(soon)* or the [`user-manual\u002F`](https:\u002F\u002Fgithub.com\u002FthClaws\u002Fuser-manual) companion repo. 24 chapters covering every feature plus 7 walkthrough case studies (static site deploy, Node.js reservation site, news-aggregation agent, etc.).\n- [Contributing](CONTRIBUTING.md) — dev setup, PR flow, commit style\n- [Changelog](CHANGELOG.md) — version history\n- [Code of Conduct](CODE_OF_CONDUCT.md) — Contributor Covenant 2.1\n- [Security](SECURITY.md) — vulnerability disclosure\n\nFor books, training, and commercial deployment, see [agentic-press.com](https:\u002F\u002Fagentic-press.com).\n\n---\n\n## License\n\nDual-licensed under either:\n\n- [MIT License](LICENSE-MIT)\n- [Apache License 2.0](LICENSE-APACHE)\n\nat your option. Contributions are accepted under the same dual license.\n\n---\n\n## About\n\nthClaws is developed by **ThaiGPT Co., Ltd.** and published under a dual MIT\u002FApache-2.0 license. The client is free and open source forever. Enterprise Edition, hosting, and support are commercial offerings — see [agentic-press.com](https:\u002F\u002Fagentic-press.com) or contact [jimmy@thaigpt.com](mailto:jimmy@thaigpt.com).\n\nBuilt in Thailand. Meant for the world.\n","thClaws是一个开源的AI代理平台，旨在为用户提供本地运行的AI工作空间。它采用Rust语言编写，支持多提供商（如Anthropic、OpenAI等），用户可以在自己的机器上运行该软件，确保数据主权。核心功能包括代码编辑与自动化、知识库搜索及团队协调，所有这些都通过一个二进制文件实现。用户可以使用自然语言指令来操作，thClaws能够读取文件、执行命令并使用工具完成任务。此外，它提供了桌面GUI、CLI和非交互模式三种接口方式，适应不同的使用场景。适用于需要高效协作、自动化处理以及基于自然语言交流的各种知识工作者，尤其是工程师、研究人员、产品经理等。",2,"2026-06-11 02:39:28","CREATED_QUERY"]