Skip to content

aarchila/Google-Antigravity-Skills-Library

 
 

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

13 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Antigravity Skills Library

Antigravity Skills Library

A collection of operational skills for AI agents and automation systems. Focused on deterministic execution, safety gates, adversarial review, and structured learning from failure. This repository provides capabilities, not autonomy, policy enforcement, or agent orchestration.

📦 Installation

This library is designed to live within the Antigravity configuration directory.

  1. Locate your specific installation directory:

    ~/.gemini/antigravity/skills/
  2. Clone this repository:

    cd ~/.gemini/antigravity/
    git clone https://github.com/Sounder25/Google-Antigravity-Skills-Library.git skills
  3. Verify: Ensure the directory structure matches: ~/.gemini/antigravity/skills/00_workspace_forensics/SKILL.md

⚙️ Prerequisites

PowerShell 5.1+ or Core 7+ is required for the underlying automation scripts.

One-time setup for your PowerShell session:

Set-ExecutionPolicy -Scope CurrentUser RemoteSigned -Force
Set-StrictMode -Version Latest
$ErrorActionPreference = "Stop"

🛡️ Robustness & Safety (Gold Standard)

Legend: ✅ Verified indicates the skill has passed deterministic tests and is behavior-frozen for v1.0.0.

17. Impasse Detector (Self-Correction)

Skill: SKILL-017: Impasse Detector ✅ Verified
Trigger: check logic, am i stuck
detect_impasse.ps1 - Analyzes transcripts for unproductive loops.

18. Pre-Action Guard (QA Gate)

Skill: SKILL-018: Pre-Action Guard ✅ Verified
Trigger: guard check, verify action
guard_check.ps1 - Validates destructive actions against plan.

19. Adversarial Reviewer (Red Team)

Skill: SKILL-019: Adversarial Reviewer ✅ Verified
Trigger: red team this
prepare_review.ps1 - Generates "Attack" prompts for self-review.

20. Failure Postmortem (Learning)

Skill: SKILL-020: Failure Postmortem ✅ Verified
Trigger: log failure
log_failure.ps1 - Structured logging of why things broke.

21. Skill Gap Identifier (Evolution)

Skill: SKILL-021: Skill Gap Identifier ✅ Verified
Trigger: new capability
propose_skill.ps1 - Scaffolds new skills from failure data.

22. State Overlay Consistency Checker

Skill: SKILL-022: State Overlay Consistency Checker ✅ Verified
Trigger: verify state continuity
Enforces state-visibility across execution contexts to prevent silent gas under-charging.

23. Paste Sanitizer

Skill: SKILL-023: Paste Sanitizer ✅ Verified
Trigger: sanitize commands
Strips prompts/output from mixed snippets to ensure safe copy-pasting.


🧠 Cognitive Skills (Phase 2 - "Superpowers")

1. Auto-Recovery Loop (Self-Healing)

Skill: SKILL-006: Error-State Recovery

2. Deterministic Planner (Architecture First)

Skill: SKILL-007: Deterministic Planner

3. Context Pruner (Focus)

Skill: SKILL-008: Context Window Pruner

4. Agent Swarm (Delegation)

Skill: SKILL-009: Agent-Swarm Spawner

5. Async Feedback (Correction)

Skill: SKILL-010: Async Feedback Loop


📚 Knowledge Skills (Phase 3)

11. Doc Parser (LLMs.txt)

Skill: SKILL-011: LLMs.txt Doc Parsing

12. Dependency Mapper (Graph)

Skill: SKILL-012: Dependency Tree Mapping

13. Environment Master (Context)

Skill: SKILL-013: Env Bash Mastery


🛠️ Housekeeping Skills

ID Name Description
000 Workspace Forensics Gold Standard Audit.
001 Run EELS Tests EVM Compliance.
002 Project Rename Safe Rename.
003 Sync Repos Multi-repo drift.
004 Clean Artifacts Disk cleanup.
005 Update Breadcrumbs Intent tracking.
014 Detect Duplicates Finds duplicate files.
015 Generate Gitignore Creates standard ignores.

🔌 Integration

MCP Connector

Skill: SKILL-016: MCP Connector
server.py - Exposes this entire library as an MCP Server.

About

Operational capabilities for high-leverage AI coding agents. Features deterministic execution, 100% tested safety gates, impasse detection, and adversarial review. Includes 'Gold Standard' meta-skills like Pre-Action Guard and Failure Postmortem to prevent loops and data loss.

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors

Languages

  • PowerShell 97.1%
  • Python 2.7%
  • C# 0.2%