app-sdk

Security and quality review of code changes. Supports branch comparison (default) and uncommitted changes modes.

Full codebase consistency scan. Checks all hand-written source files against project coding rules for style, pattern, and quality violations. Excludes auto-generated files and node_modules. Use periodically (e.g. before a release or after onboarding new team members) to catch drift from project standards.

Pre-PR quality check with adaptive agent selection, code simplification with user approval, and optimized validation. Scales review depth by diff size.

Post-implementation side effect analysis. Checks consumer impact, shared state, UI components, function signatures, and layout cascade for unintended regressions.

Show concise summary of all changes on current branch vs parent branch. Outputs grouped bullet-point list.

Create React UI components. Use when building new components, forms, dialogs, data tables, or implementing UI features. Guides through component creation workflow from planning to verification.

Debug UI issues using Chrome browser automation. Use when investigating visual bugs, console errors, network failures, or unexpected component behavior. Requires --chrome flag.

Full codebase consistency scan. Checks all hand-written source files against project coding rules for style, pattern, and quality violations. Excludes auto-generated files and node_modules. Use periodically (e.g. before a release or after onboarding new team members) to catch drift from project standards.

Extract reusable patterns from current session. Saves non-trivial debugging techniques, workarounds, and project conventions as skill files.

Clean Tailwind CSS classnames in modified files. Converts arbitrary values to standard classes, ensures cn() usage, and removes redundant styles.

Generate PR summary by comparing code differences with parent branch. Uses PR template and outputs to ./report/PR_SUMMARY.MD.

Pre-PR quality check with adaptive agent selection, code simplification with user approval, and optimized validation. Scales review depth by diff size.

Review mistakes and update rules. Use after debugging sessions, user corrections, or when anti-patterns are found. Updates .claude/rules/*.md to prevent recurrence.

Post-implementation side effect analysis. Checks consumer impact, shared state, UI components, function signatures, and layout cascade for unintended regressions.

Create agent teams for efficient, high-quality task delivery. Spawns specialized teammates (UX, Architecture, Web Researcher, Devil's Advocate, Security, Performance, Code Reviewer, QA) to collaboratively deliver features with precision and quality.

Todo-driven development for long-running AI tasks. Externalizes task state to todo.md to prevent AI goal drift. Use /todo init to create a structured task plan, /todo check for mid-task progress checkpoints, /todo score for final weighted scoring, /todo clear to archive and reset.