tzachbon/smart-ralph

Spec-driven development with smart compaction. Claude Code plugin combining Ralph Wiggum loop with structured specification workflow.

What it solves

Smart Ralph is a spec-driven development framework for AI coding agents (specifically Claude Code and Codex). It prevents agents from jumping straight into coding without a plan, instead forcing a structured process of research, requirements gathering, and technical design before any implementation begins. This ensures that complex features are well-thought-out and can be resumed or audited across multiple sessions.

How it works

The project implements a multi-agent workflow where different specialized agents handle specific phases of development:

  1. Triage: Splits large goals into smaller, dependency-aware specs (Epics).
  2. Research: Analyzes the existing codebase and searches the web for feasibility.
  3. Requirements: Defines user stories and acceptance criteria.
  4. Prototype: (Optional) Tests a specific design question in a disposable environment.
  5. Design: Defines the architecture and technical trade-offs.
  6. Tasks: Breaks the design into a sequence of executable tasks.
  7. Execution: Implements the tasks one by one, using fresh context for each task and applying quality gates (POC, refactoring, testing, and linting).

It also includes a codebase indexing feature that scans projects to create searchable component specs, helping the research agents find existing code more efficiently.

Who it’s for

Developers using Claude Code or OpenAI Codex who want a more disciplined, architectural approach to AI-assisted feature development rather than simple prompt-based code generation.

Highlights

  • Spec-Driven Workflow: Forces a sequence of research $\rightarrow$ requirements $\rightarrow$ design $\rightarrow$ tasks $\rightarrow$ implementation.
  • State Persistence: Records progress in .progress.md and tasks.md, allowing work to be resumed after a session stop.
  • Codebase Indexing: Automatically scans controllers, services, and models to create a searchable index for the AI.
  • Multi-Agent Specialization: Uses distinct roles (e.g., triage-analyst, product-manager, architect-reviewer) for different phases.
  • Flexible Execution: Supports both a detailed "normal" mode with approval checkpoints and a "quick" mode for faster iteration.

Related

  • Project
    rsmdt/the-startup

    A multi-agent AI framework for Claude Code that implements a spec-driven development workflow to turn ideas into shipped code via structured planning and parallel specialist agents.

  • Project
    umputun/ralphex

    A standalone CLI tool that orchestrates Claude Code or codex to autonomously execute implementation plans, featuring a multi-phase AI code review pipeline.

  • Project
    DannyMac180/fable-advisor

    A Claude Code plugin that implements an architect pattern, orchestrating tasks between Fable 5.1 and GPT-5.6 models to optimize cost and quality through cross-vendor review.

  • Project
    umputun/cc-thingz

    cc‑thingz is a marketplace of Claude Code plugins that add higher‑level developer workflows—brainstorming, PR review, structured planning, release automation, and assorted thinking tools. Install via Claude’s `/plugin` command, configure a few JSON hooks, and invoke the new slash commands (e.g., `/brainstorm:do`, `/review:pr`, `/planning:make`). The plugins use Claude sub‑agents, hooks, and editor overlays to automate planning, code review, and even autonomous plan execution, making Claude Code a more powerful coding assistant.

  • Project
    CloudAI-X/claude-workflow-v2

    A universal workflow plugin for Claude Code and other AI agents that provides specialized agents, slash commands, and automation hooks to streamline software development.