Codelab: Setting Up AGENTS.md, Skills, and Rules for Antigravity
Overview
[TODO: 2-3 sentence intro — what Antigravity is, why AGENTS.md/skills/rules matter, and what the reader will walk away being able to do.]
Prerequisites
[TODO: e.g. Antigravity installed / access granted] [TODO: e.g. A sample project or repo to work in] [TODO: Any accounts, CLI tools, or API keys needed] ---
Step 1: What You'll Build
[TODO: Briefly describe the end state — a project configured with a working AGENTS.md, at least one custom skill, and a rule set — and why that combination matters for agentic workflows.] ---
Step 2: Create Your AGENTS.md
2.1 Purpose of AGENTS.md [TODO: Explain what AGENTS.md does in Antigravity — e.g. how it gives the agent context about the project, conventions, and constraints.] 2.2 Scaffold the file [TODO: example AGENTS.md skeleton / starter template] 2.3 Fill in project-specific context [TODO: Walk through the sections a reader should fill in — project overview, tech stack, coding conventions, do's and don'ts, etc.] [TODO: Callout — any gotchas or best practices worth highlighting] ---
Step 3: Build a Custom Skill
3.1 What a skill is [TODO: Explain the concept of a skill in Antigravity — a packaged, reusable set of instructions/tools for a specific task.] 3.2 Create the skill file [TODO: example skill file structure/content] 3.3 Test the skill [TODO: Steps to trigger and verify the skill works as expected.] ---
Step 4: Configure Rules
4.1 What rules control [TODO: Explain what "rules" govern in Antigravity — guardrails, formatting requirements, tool permissions, etc.] 4.2 Write your first rule set [TODO: example rules file/config] 4.3 Verify enforcement [TODO: How to confirm the rules are actually being applied — example prompt/response before and after.] ---
Step 5: Put It All Together
[TODO: Run a realistic end-to-end task that exercises AGENTS.md context, the custom skill, and the rules together. Show the expected output.] ---
Summary
[TODO: Recap what was configured and why it matters.] What you learned [TODO: Key takeaway 1] [TODO: Key takeaway 2] [TODO: Key takeaway 3] Next steps [TODO: Suggested follow-up — e.g. adding more skills, sharing AGENTS.md across a team, etc.]