Can I choose which code review findings to fix?

Absolutely. After a code review, the skill presents findings grouped by severity (Critical, High, Medium, Low). You can interactively select which levels to address, and it will spawn a subagent to apply those specific fixes.

Does this skill require a pre-existing technical plan?

Yes, it is designed to follow a structured plan document. It works best when paired with the 'iterative:tech-planning' skill to ensure all dependencies and test scenarios are defined before coding starts.

How does the skill handle failing tests?

The skill follows a strict 'no-commit-on-failure' policy. It runs tests after every change and includes verification gates that prevent the workflow from proceeding if tests fail or if required test files are missing.

What is workspace isolation in this context?

The skill manages your Git environment by offering to create dedicated worktrees or feature branches. This keeps your main development branch clean and ensures that the AI-driven implementation happens in a controlled, isolated space.

Iterative Implementation

Name: Iterative Implementation
Author: tmchow

bytmchow

•

Productivity & Workflow

Executes structured technical plans with dependency-aware task batching, test-driven development, and automated multi-stage code reviews.

The Iterative Implementation skill automates the transition from planning to production by managing the entire development lifecycle within Claude Code. It interprets technical plan documents to create a dependency-aware task graph, setting up isolated Git workspaces to ensure clean implementation. By enforcing strict Test-Driven Development (TDD) and continuous testing gates, the skill ensures that every commit is verified. It features a sophisticated multi-tier review system—performing section-level reviews during build-out and a final simplification pass—allowing users to interactively accept and apply fixes based on severity before generating a pull request.

Key Features

01Workspace isolation using Git worktrees or feature branches to prevent environment pollution.

02Dependency-aware task batching for efficient, sequential execution of complex plans.

03Automated code simplification pass and PR preparation including behavior-preserving refactors.

0420 GitHub stars

05Tiered code reviews with interactive severity-based acceptance (Critical, High, Medium, Low).

06Enforced Test-Driven Development (TDD) with automated verification gates for every feature.

Use Cases

01Executing complex, multi-step features from a structured technical design document.

02Managing large-scale refactors that require isolated environments and rigorous testing gates.

03Automating the standard 'Build-Test-Review-Commit' cycle for team-wide engineering consistency.

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add tmchow/tmc-marketplace implementing

For use in Claude.ai and ChatGPT

Download Skill