The visual management layer for AI Agent skills, workflows, and rules.
English | 繁體中文
Antigravity was born on Christmas 2025 as a Python CLI tool for managing AI agent skills. Within four months, it evolved into a full VS Code extension with 4,300+ downloads, a force-directed topology graph, and a conversation archive system that could decrypt Gemini's binary .pb logs into searchable Markdown.
It was built by the ChronicleCore A1 Expert System — a structured team of 38 specialized AI agents, each with defined roles, permissions, and persistent memory, collaborating under a unified governance protocol. This extension was the first visual proof that such a system could produce real, shippable software.
The A1 system has since evolved beyond the Antigravity architecture. This codebase is now fully open-sourced under MIT — both as a working tool for the community and as a historical artifact of a new approach to AI-assisted software development.
For the architectural framework behind this project, see ChronicleCore-Architecture.
| Date | Version | Milestone |
|---|---|---|
| 2025-12-25 | v0.1.0 |
Origin — Python CLI based on skills-cli by KC Chien |
| 2026-01-21 | v1.0.0 |
Antigravity born — transformed into VS Code Extension |
| 2026-01-22 | v1.0.3~1.0.7 |
Five versions in a single day of explosive development |
| 2026-01-25 | v1.2.6 |
First stable release |
| 2026-02-25 | v2.0.0 |
Major rewrite — Star Map topology (d3-force), Activity Bar, conversation archive |
| 2026-03-05 | v2.1.0 |
Archive stabilizer + esbuild migration (VSIX 96MB → 2.3MB) |
| 2026-03-09 | — | Final update before architecture migration |
| 2026-04 | — | A1 Expert System migrates to next-gen architecture. Antigravity completes its mission |
| 2026-04-20 | — | MIT open-sourced — all users upgraded to Golden Supporter |
| Feature | Description |
|---|---|
| Star Map Topology | Dynamic force-directed graph showing relationships between agents, skills, and data (d3-force) |
| Skills Dashboard | View and manage all skills in .agent/skills/ |
| Workflow Manager | Organize and execute workflows from .agent/workflows/ |
| Rules Inspector | View and edit local (.agent/rules/) & global (GEMINI.md) rules |
| Conversation Archive | Decrypt .pb protobuf logs into searchable Markdown |
| Dynamic Status Bar | Real-time skill and workflow count in the IDE |
Navigate knowledge, not folders. The Star Map visualizes complex connections between your agents, skills, and data.
Decrypt and parse Gemini's .pb conversation files into human-readable, full-text searchable Markdown.
High-level overview of skill, workflow, and rule counts — mission control for your AI agent environment.
Inspect skill metadata, behavioral constraints, persona configurations, and synchronized memory.
Trigger complex multi-agent rituals with detailed SOP files and automation chains.
Search "Antigravity Skills Chronicle" in VS Code Extensions, or:
ext install ChronicleCore.antigravity-skills-chronicle
git clone https://github.com/Zaious/Antigravity-Skills-Chronicle.git
cd Antigravity-Skills-Chronicle
npm install
cd web && npm install && cd ..
npm run bundleLocal (per project)
- Skills:
.agent/skills/ - Workflows:
.agent/workflows/ - Rules:
.agent/rules/
Global (user profile)
- Global Rules:
~/.gemini/antigravity/GEMINI.md - Chronicle Exports:
~/.gemini/antigravity/.search_cache/
.agent/ # Example agent data (skills, workflows, rules)
│ # Clone and open in VS Code to see the dashboard in action
│
src/ # VS Code Extension backend (Node.js/TypeScript)
├── extension.ts # Main entry point
├── core/ # History extraction, indexing
├── providers/ # VS Code TreeView providers
└── remote-bridge/ # Discord & Telegram bot integration
web/ # Webview frontend (React + Vite + Tailwind)
└── src/
├── App.tsx # Dashboard UI
├── components/ # Reusable components
├── hooks/ # React hooks
└── types/ # TypeScript type definitions
This project was developed by a team of 5 AI agents (selected from a pool of 38) operating under a structured governance protocol. Each agent had defined roles, permission tiers, and accountability rules.
See AGENTS.md for the original team roster and collaboration protocols, or the English annotated version for a detailed explanation of how multi-agent software development works.
We welcome contributions! See CONTRIBUTING.md for guidelines.
This project is community-maintained. ChronicleCore reviews and publishes approved PRs to the VS Code Marketplace.
- Original CLI foundation: skills-cli by KC Chien (MIT License)
- Extension development: Zaious / ChronicleCore A1 Expert System
MIT — see THIRD-PARTY-NOTICES.md for upstream attributions.
Built with the ChronicleCore A1 Expert System — 38 specialized AI agents, one unified architecture.
Made by Zaious / ChronicleCore / 編年史記工作室




