{"id":33996104,"url":"https://github.com/study8677/antigravity-workspace-template","last_synced_at":"2026-03-04T06:17:58.450Z","repository":{"id":325058284,"uuid":"1099636778","full_name":"study8677/antigravity-workspace-template","owner":"study8677","description":"🪐 The ultimate starter kit for Google Antigravity IDE. Optimized for Gemini 3 Agentic Workflows, \"Deep Think\" mode, and auto-configuring .cursorrules.","archived":false,"fork":false,"pushed_at":"2026-02-27T02:11:34.000Z","size":966,"stargazers_count":928,"open_issues_count":2,"forks_count":187,"subscribers_count":18,"default_branch":"main","last_synced_at":"2026-02-27T09:41:05.030Z","etag":null,"topics":["agentic-ai","deep-think","gemini3pro","google-antigravity","python-template"],"latest_commit_sha":null,"homepage":"https://github.com/study8677/antigravity-workspace-template","language":"Python","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":"mit","status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/study8677.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":null,"funding":null,"license":"LICENSE","code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":null,"support":null,"governance":null,"roadmap":null,"authors":null,"dei":null,"publiccode":null,"codemeta":null,"zenodo":null,"notice":null,"maintainers":null,"copyright":null,"agents":"AGENTS.md","dco":null,"cla":null}},"created_at":"2025-11-19T08:51:14.000Z","updated_at":"2026-02-27T08:10:57.000Z","dependencies_parsed_at":null,"dependency_job_id":null,"html_url":"https://github.com/study8677/antigravity-workspace-template","commit_stats":null,"previous_names":["study8677/antigravity-workspace-template"],"tags_count":1,"template":false,"template_full_name":null,"purl":"pkg:github/study8677/antigravity-workspace-template","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/study8677%2Fantigravity-workspace-template","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/study8677%2Fantigravity-workspace-template/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/study8677%2Fantigravity-workspace-template/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/study8677%2Fantigravity-workspace-template/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/study8677","download_url":"https://codeload.github.com/study8677/antigravity-workspace-template/tar.gz/refs/heads/main","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/study8677%2Fantigravity-workspace-template/sbom","scorecard":null,"host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":286080680,"owners_count":30074017,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2026-03-04T05:31:57.858Z","status":"ssl_error","status_checked_at":"2026-03-04T05:31:38.462Z","response_time":59,"last_error":"SSL_connect returned=1 errno=0 peeraddr=140.82.121.5:443 state=error: unexpected eof while reading","robots_txt_status":"success","robots_txt_updated_at":"2025-07-24T06:49:26.215Z","robots_txt_url":"https://github.com/robots.txt","online":false,"can_crawl_api":true,"host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":["agentic-ai","deep-think","gemini3pro","google-antigravity","python-template"],"created_at":"2025-12-13T08:00:40.444Z","updated_at":"2026-03-04T06:17:58.430Z","avatar_url":"https://github.com/study8677.png","language":"Python","funding_links":[],"categories":["CLIs","Python"],"sub_categories":["Usage"],"readme":"# 🪐 Google Antigravity Workspace Template\n\n**Production-grade starter kit for autonomous AI agents on Google Antigravity.**\n\nLanguage: [English](/docs/en/) | [中文（仓库主页）](README_CN.md) | [中文文档](/docs/zh/) | [Español](/docs/es/)\n\n![License](https://img.shields.io/badge/License-MIT-green)\n![Gemini](https://img.shields.io/badge/AI-Gemini_2.0_Flash-blue)\n![Architecture](https://img.shields.io/badge/Architecture-Event_Driven-purple)\n![Memory](https://img.shields.io/badge/Context-Infinite-orange)\n\n## 🌟 Project Intent\n\nIn a world full of AI IDEs, I want enterprise-grade architecture to be as simple as **Clone → Rename → Prompt**.\n\nThis project leverages IDE context awareness (via `.cursorrules` and `.antigravity/rules.md`) to pre-embed a complete **cognitive architecture** in the repo.\n\nWhen you open this project, your IDE stops being just an editor—it becomes an **industry-savvy architect**.\n\n**First principles:**\n\n- Minimize repetition: the repo should encode defaults so setup is nearly zero.\n- Make intent explicit: capture architecture, context, and workflows in files, not tribal knowledge.\n- Treat the IDE as a teammate: contextual rules turn the editor into a proactive architect, not a passive tool.\n\n### Why do we need a thinking scaffold?\n\nWhile building with Google Antigravity or Cursor, I found a pain point:\n\n**The IDE and models are powerful, but the empty project is too weak.**\n\nEvery new project repeats the same boring setup:\n\n- \"Should my code live in `src` or `app`?\"\n- \"How do I define utilities so Gemini recognizes them?\"\n- \"How do I help the AI remember prior context?\"\n\nThis repetition wastes creative energy. My ideal workflow is: **after a git clone, the IDE already knows what to do.**\n\nSo I built this project: **Antigravity Workspace Template**.\n\n## ⚡ Quick Start\n\n### Automated Installation (Recommended)\n\n**Linux / macOS:**\n```bash\n# 1. Clone the template\ngit clone https://github.com/study8677/antigravity-workspace-template.git my-project\ncd my-project\n\n# 2. Run the installer\nchmod +x install.sh\n./install.sh\n\n# 3. Configure your API keys\nnano .env\n\n# 4. Run the agent\nsource venv/bin/activate\npython src/agent.py\n```\n\n**Windows:**\n```cmd\n# 1. Clone the template\ngit clone https://github.com/study8677/antigravity-workspace-template.git my-project\ncd my-project\n\n# 2. Run the installer\ninstall.bat\n\n# 3. Configure your API keys (notepad .env)\n\n# 4. Run the agent\npython src/agent.py\n```\n\n### Manual Installation\n\n```bash\n# 1. Clone the template\ngit clone https://github.com/study8677/antigravity-workspace-template.git my-project\ncd my-project\n\n# 2. Create virtual environment\npython3 -m venv venv\nsource venv/bin/activate  # On Windows: venv\\Scripts\\activate\n\n# 3. Install dependencies\npip install -r requirements.txt\n\n# 4. Configure your API keys\ncp .env.example .env  # (if available) or create .env manually\nnano .env\n\n# 5. Run the agent\npython src/agent.py\n```\n\n**That's it!** The IDE auto-loads configuration via `.cursorrules` + `.antigravity/rules.md`. You're ready to prompt.\n\n## 🎯 What Is This?\n\nThis is **not** another LangChain wrapper. It's a minimal, transparent workspace for building AI agents that:\n\n- 🧠 Have infinite memory (recursive summarization)\n- 🛠️ Auto-discover tools from `src/tools/`\n- 📚 Auto-inject context from `.context/`\n- 🔌 Connect to MCP servers seamlessly\n- 🤖 Coordinate multiple specialist agents\n- 📦 Save outputs as artifacts (plans, logs, evidence)\n\n**Clone → Rename → Prompt. That's the workflow.**\n\n## 🚀 Key Features\n\n| Feature | Description |\n|---------|-------------|\n| 🧠 **Infinite Memory** | Recursive summarization compresses context automatically |\n| 🧠 **True Thinking** | \"Deep Think\" step using Chain-of-Thought prompts before acting |\n| 🎓 **Skills System** | Modular capabilities as folders (`src/skills/`) with auto-loading (includes `agent-repo-init`) |\n| 🛠️ **Universal Tools** | Drop Python functions in `src/tools/` → auto-discovered |\n| 📚 **Auto Context** | Add files to `.context/` → auto-injected into prompts |\n| 🔌 **MCP Support** | Connect GitHub, databases, filesystems, custom servers |\n| 🤖 **Swarm Agents** | Multi-agent orchestration with Router-Worker pattern |\n| ⚡ **Gemini Native** | Optimized for Gemini 2.0 Flash |\n| 🌐 **LLM Agnostic** | Use OpenAI, Azure, Ollama, or any OpenAI-compatible API |\n| 📂 **Artifact-First** | Convention-first workflow for storing plans, logs, and evidence in `artifacts/` |\n| 🔒 **Sandbox Execution** | Configurable code execution environments (local by default) |\n\n## 📚 Documentation\n\n**Full documentation available in `/docs/en/`:**\n\n- **[Quick Start](docs/en/QUICK_START.md)** — Installation \u0026 deployment\n- **[Philosophy](docs/en/PHILOSOPHY.md)** — Core concepts \u0026 architecture\n- **[Zero-Config](docs/en/ZERO_CONFIG.md)** — Auto tool \u0026 context loading\n- **[MCP Integration](docs/en/MCP_INTEGRATION.md)** — External tool connectivity\n- **[Swarm Protocol](docs/en/SWARM_PROTOCOL.md)** — Multi-agent coordination\n- **[Roadmap](docs/en/ROADMAP.md)** — Future phases \u0026 vision\n\n### Sandbox Configuration (Zero-Config by default)\n\nThe sandbox lets the agent execute generated Python code safely and consistently. It defaults to a local subprocess with isolation and limits.\n\n- `SANDBOX_TYPE`: `local` (default) | `microsandbox` (opt-in) | `e2b` (future)\n- `SANDBOX_TIMEOUT_SEC`: maximum execution time in seconds (default `30`)\n- `SANDBOX_MAX_OUTPUT_KB`: truncate stdout/stderr to limit size (default `10`)\n\nMicrosandbox (opt-in) extra variables:\n- `MSB_SERVER_URL` (default `http://127.0.0.1:5555`)\n- `MSB_API_KEY` (optional)\n- `MSB_IMAGE` (default `microsandbox/python`)\n- `MSB_CPU_LIMIT` (default `1.0`)\n- `MSB_MEMORY_MB` (default `512`)\n\nExample:\n\n```bash\nexport SANDBOX_TYPE=local\nexport SANDBOX_TIMEOUT_SEC=30\nexport SANDBOX_MAX_OUTPUT_KB=10\n# Microsandbox mode\n# msb server start --dev\n# export SANDBOX_TYPE=microsandbox\n# export MSB_SERVER_URL=http://127.0.0.1:5555\n# export MSB_IMAGE=microsandbox/python\n```\n\n## 🏗️ Project Structure\n\n```\nsrc/\n├── agent.py           # Main agent loop\n├── memory.py          # JSON memory manager\n├── mcp_client.py      # MCP integration\n├── swarm.py           # Multi-agent orchestration\n├── agents/            # Specialist agents\n├── tools/             # Your custom tools\n└── skills/            # Modular skills (Zero-Config)\n\n.context/             # Knowledge base (auto-injected)\n.antigravity/         # Antigravity rules\nartifacts/            # Outputs \u0026 evidence\n```\n\n## 💡 Example: Build a Tool in 30 Seconds\n\n```python\n# src/tools/my_tool.py\ndef analyze_sentiment(text: str) -\u003e str:\n    \"\"\"Analyzes the sentiment of given text.\"\"\"\n    return \"positive\" if len(text) \u003e 10 else \"neutral\"\n```\n\n**Restart agent.** Done! The tool is now available.\n\n## 🎓 Example: Initialize a New Repo with Skill\n\nThe built-in `agent-repo-init` skill supports two modes:\n- `quick`: minimal clean scaffold\n- `full`: scaffold + runtime profile defaults (`.env`, mission, context profile, init report)\n\nYou can run the portable script at `skills/agent-repo-init/scripts/init_project.py`:\n\n```text\npython skills/agent-repo-init/scripts/init_project.py \\\n  --project-name my-new-agent \\\n  --destination-root /absolute/path/for/new/projects \\\n  --mode quick\n```\n\n`full` mode example adds profile defaults:\n\n```text\npython skills/agent-repo-init/scripts/init_project.py \\\n  --project-name my-new-agent \\\n  --destination-root /absolute/path/for/new/projects \\\n  --mode full --llm-provider openai --enable-mcp --disable-swarm --sandbox-runtime microsandbox --init-git\n```\n\n## 🔌 MCP Integration\n\nConnect to external tools:\n\n```json\n{\n  \"servers\": [\n    {\n      \"name\": \"github\",\n      \"transport\": \"stdio\",\n      \"command\": \"npx\",\n      \"args\": [\"-y\", \"@modelcontextprotocol/server-github\"],\n      \"enabled\": true\n    }\n  ]\n}\n```\n\nAgent automatically discovers and uses all MCP tools.\n\n## 🤖 Multi-Agent Swarm\n\nDecompose complex tasks:\n\n```python\nfrom src.swarm import SwarmOrchestrator\n\nswarm = SwarmOrchestrator()\nresult = swarm.execute(\"Build and review a calculator\")\n```\n\nThe swarm automatically:\n- 📤 Routes to Coder, Reviewer, Researcher agents\n- 🧩 Synthesizes results\n- 📂 Exposes message logs via `get_message_log()` for inspection\n\n## ✅ What's Complete\n\n- ✅ Phase 1-7: Foundation, DevOps, Memory, Tools, Swarm, Discovery\n- ✅ Phase 8: MCP Integration (fully implemented)\n- 🚀 Phase 9: Enterprise Core (in progress)\n\n## 🆕 Recent Updates\n\n- 2026-03-02: Migrated sandbox runtime from Docker to Microsandbox.\n- Added `microsandbox` backend with `SANDBOX_TYPE=microsandbox` (default remains `local`).\n- Removed Docker runtime implementation and Docker sandbox test suite.\n- Updated sandbox env/config keys to `MSB_SERVER_URL`, `MSB_API_KEY`, `MSB_IMAGE`, `MSB_CPU_LIMIT`, `MSB_MEMORY_MB`.\n- Breaking change: `agent-repo-init` replaced `enable_docker` with `sandbox_runtime` (`local` | `microsandbox`).\n- Updated init script usage to `--sandbox-runtime microsandbox` (replaces `--enable-docker`).\n- Added **True Thinking**: The agent now performs a real \"Deep Think\" step (Chain-of-Thought) before every action, generating a structured plan.\n- Added **Skills System**: New `src/skills/` directory allows for modular, folder-based agent capabilities (Docs + Code).\n- Added **agent-repo-init skill**: Initialize a clean, reusable repository from this template via `init_agent_repo`.\n- Added local OpenAI-compatible backend support (e.g., Ollama) when no Google API key is provided.\n- Fixed `.env` loading so runs from the `src/` folder still read the project-root config.\n- CLI entrypoints (`agent.py` and `src/agent.py`) now accept tasks via arguments `AGENT_TASK`.\n\nSee [Roadmap](docs/en/ROADMAP.md) for details.\n\n## 🤝 Contributing\n\nIdeas are contributions too! Open an [issue](https://github.com/study8677/antigravity-workspace-template/issues) to:\n- Report bugs\n- Suggest features\n- Propose architecture (Phase 9)\n\nOr submit a PR to improve docs or code.\n\n## 👥 Contributors\n\n- [@devalexanderdaza](https://github.com/devalexanderdaza) — First contributor. Implemented demo tools, enhanced agent functionality, proposed the \"Agent OS\" roadmap, and completed MCP integration.\n- [@Subham-KRLX](https://github.com/Subham-KRLX) — Added dynamic tools and context loading (Fixes #4) and the multi-agent cluster protocol (Fixes #6).\n\n## ⭐ Star History\n\n[![Star History Chart](https://api.star-history.com/svg?repos=study8677/antigravity-workspace-template\u0026type=Date)](https://star-history.com/#study8677/antigravity-workspace-template\u0026Date)\n\n## 📄 License\n\nMIT License. See [LICENSE](LICENSE) for details.\n\n---\n\n**[Explore Full Documentation →](docs/en/)**\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fstudy8677%2Fantigravity-workspace-template","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fstudy8677%2Fantigravity-workspace-template","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fstudy8677%2Fantigravity-workspace-template/lists"}