Run a comprehensive code review using available AI tools in parallel. Analyzes architecture, code quality, security, and test coverage. Dynamically detects and leverages available MCP servers for multi-perspective review. Use after implementation or refactoring, before creating a PR.
Run a comprehensive code review using available AI tools in parallel. Analyzes architecture, code quality, security, and test coverage. Dynamically detects and leverages available MCP servers for multi-perspective review. Use when reviewing code changes, before merging PRs, or for quality audits.
Run iterative refactoring on changed files (minimum 3 rounds). Round 1: structural improvements (DRY, method extraction). Round 2: readability (naming, nesting reduction). Round 3: convention compliance. Continues until no improvements found. Generates a before/after report. Use after implementation, before code review.
Run the full development workflow end-to-end: specify → implement → refactor → code-review → pr-summary. Pauses after specify for user confirmation before proceeding. Pass a feature description to start from specification, or a feature name to resume from implementation.
Start implementation based on a specification document (order.md). Processes tasks from todo.md sequentially. Supports TDD (Red-Green-Refactor) or standard implementation workflow. Auto-detects project's test framework, lint tools, and coding conventions. Use after /specify.
Auto-generate a pull request description from git changes. Analyzes commits, diffs, and code structure to produce a comprehensive PR summary. Automatically detects and follows your project's PR template if present. Use after completing implementation, before creating a PR.
Generate a specification document and task list for a feature, bugfix, improvement, spike, or chore. Automatically detects the scenario type, analyzes existing codebase patterns, and produces an actionable spec (order.md) with a task checklist (todo.md). Use before starting implementation.
Implement a Flutter screen or widget from a Figma URL with design-perfect accuracy. Extracts exact numerical values via Figma MCP and translates them 1-to-1 into Flutter code. No guessing, no rounding, no approximation.