Understanding the output

What gets generated

The CLI writes adapter-native files directly into your project. You get only the files for the IDEs you selected — nothing else is created.

Commit these files like any other project config. The CLI creates AI instructions and adapter files, not runnable application code. Review and edit the generated content before treating it as authoritative.

Adapter	What gets written
Cursor	.cursor/rules/index.mdc (agent + context) · .cursor/rules/<rule>.mdc (13 rules, per-rule globs) · .cursor/skills/<stage>/SKILL.md (7 lifecycle stages)
Claude Code	.claude/rules/<rule>.md (13 rules) · .claude/commands/<stage>.md (7 lifecycle stages) · CLAUDE.md (load-when pointer index)
VS Code Copilot	.github/copilot-instructions.md (agent + all rules + all lifecycle stages merged into one file)
Windsurf	.windsurf/rules/<rule>.md (13 rules) · .windsurf/rules/lifecycle-<stage>.md (7 stages) · .windsurfrules (agent + context)
Antigravity	.agents/workflows/<stage>.md (7 lifecycle stages)

Select one or more adapters during the interactive prompt. Only the files for selected adapters are written.

Rules

Rule file	What it controls
architecture.md	Component layout, rendering strategy, domain structure
components.md	Component design, props contracts, composition patterns
styling-accessibility.md	Styling approach, a11y requirements, keyboard/focus
state-and-data-fetching.md	State strategy, query patterns, cache policy
routing.md	Route structure, navigation, meta-framework conventions
forms-validation.md	Form strategy, validation, submission patterns
performance-and-testing.md	CWV, bundle size, testing strategy and framework idioms
errors-logging.md	Error handling, logging, observability
security.md	Auth, CSRF, secrets handling
seo-meta.md	Meta tags, structured data, SSR/SSG conventions
environment.md	Config, env files, CI flags
git-conventions.md	Branches, commits, PR scope
pre-commit.md	Build / lint / test gates before commit

Lifecycle stages

The 7 lifecycle stages are written in adapter-native paths (see table above):

• think — understand goal, stack, constraints, and existing code before planning
• plan — name affected routes, components, state/data boundaries, risks, and tests
• build — implement only the approved frontend scope
• review — check correctness, accessibility, performance, maintainability
• test — run or define validation proportional to risk
• ship — summarize changes, validation, release notes, and rollback notes
• reflect — capture template gaps and follow-up tasks

Per-adapter output

Cursor

.cursor/rules/index.mdc           # agent identity + context (alwaysApply: true)
.cursor/rules/<rule>.mdc          # one file per rule, globs where relevant
.cursor/skills/think/
.cursor/skills/plan/
.cursor/skills/build/
.cursor/skills/review/
.cursor/skills/test/
.cursor/skills/ship/
.cursor/skills/reflect/

Claude Code

CLAUDE.md                         # slim pointer index — lists all rules and commands with load-when paths
.claude/rules/<rule>.md           # one file per rule (loaded on demand)
.claude/commands/think.md
.claude/commands/plan.md
.claude/commands/build.md
.claude/commands/review.md
.claude/commands/test.md
.claude/commands/ship.md
.claude/commands/reflect.md

VS Code Copilot

.github/copilot-instructions.md   # agent identity + all rules + all lifecycle stages merged

Windsurf

.windsurfrules                         # agent identity (slim, always-on)
.windsurf/rules/<rule>.md              # one file per rule
.windsurf/rules/lifecycle-think.md
.windsurf/rules/lifecycle-plan.md
.windsurf/rules/lifecycle-build.md
.windsurf/rules/lifecycle-review.md
.windsurf/rules/lifecycle-test.md
.windsurf/rules/lifecycle-ship.md
.windsurf/rules/lifecycle-reflect.md

Antigravity

.agents/workflows/think.md
.agents/workflows/plan.md
.agents/workflows/build.md
.agents/workflows/review.md
.agents/workflows/test.md
.agents/workflows/ship.md
.agents/workflows/reflect.md

Next: Recommended workflow.