Working with Cursor agents

This repository includes Cursor-specific configuration so agents understand the icon pipeline, design contract, and deployment setup.

For humans

1. Open the repo in Cursor.
2. Enable Project Rules (.cursor/rules/).
3. Read the root AGENTS.md handbook — it is the single source of truth for layout and commands.
4. Optional MCP: copy .cursor/mcp.example.json to .cursor/mcp.json and set tokens in your environment (do not commit secrets).

What is configured

Feature	Location
Agent handbook	AGENTS.md
File-scoped rules	.cursor/rules/*.mdc
Workflows (skills)	.cursor/skills/muralize-icons, .cursor/skills/add-icon, .cursor/skills/build-and-verify
Cursor folder readme	.cursor/README.md

Example prompts

• "Add a bookmark-filled icon following design guidelines."
• "Run the build-and-verify skill and fix any lint errors."
• "Update IconGrid to filter by category transport."
• "Why did lint:icons fail?"

Icon contributions

Agents should follow the same steps as human contributors:

1. design/guidelines.md
2. Update scripts/mural-icon-paths.mjs + icons/{name}.json
3. pnpm icons:mural → pnpm lint:icons → pnpm build
4. PR with 16 / 24 / 32px previews

See Contributing.

Verification

Agents verifying changes should run the build-and-verify skill (.cursor/skills/build-and-verify/SKILL.md): lint icons, build, typecheck, docs build.