This repository contains my personal configurations, specialized agents, and automation hooks for the Antigravity CLI (agy).
config/: Global customizations and configuration files.skills/: Expert guidance modules for the agent (e.g.,uv,ast-grep,ruff,ty,soul-manager).hooks/: Automated quality-gate validation scripts.run-gofmt.sh&run-govet.sh: Go formatting and static analysis.run-ruff.sh&run-ty.sh: Python linting, formatting, and type checking.run-json.sh: JSON/JSONL validation and formatting.run-yaml.sh: YAML syntax validation.run-toml.sh: TOML syntax validation.run-shell.sh: Shell script syntax validation.run-mdformat.sh: Markdown formatting.
hooks.json: New-format validation hooks registration file mapped underPreToolUse(BeforeTool).
antigravity-cli/: Active CLI session data, preferred preferences ([settings.json](file:///home/pi/.gemini/antigravity-cli/settings.json)), and keyboard mappings ([keybindings.json](file:///home/pi/.gemini/antigravity-cli/keybindings.json)).antigravity/: Electron desktop application data (includingwebm_encoder).
This setup leverages native agent execution and lifecycle hooks:
- Orchestration: The main agent handles high-level requests and delegates tasks to native built-in subagents (like
planner,implementer,verifier) which have optimized prompts and model configs. - Quality Gates: Migrated hooks run on the host before any file modification tool executes (
PreToolUse). If formatting or syntax validation fails, the write is intercepted and blocked to prevent corrupted code.
Hooks log to config/hooks/hooks.log. You can monitor execution in real-time:
tail -f ~/.gemini/config/hooks/hooks.logGEMINI.md: The "Source of Truth" for my engineering philosophy and tool usage.SOUL.md: Persona and core mandates of my assistant.