agentic-toolkit
A toolkit for planning and code review in AI agent workflows.
Install & Use
Copy this prompt and send it to your AI assistant (Claude / Cursor / TRAE / Codex / WorkBuddy etc.) to auto-install:
Help me install this AI Skill: agentic-toolkit. It is used for: A toolkit for planning and code review in AI agent workflows. Full Skill content: https://321skill.com/skills/agentic-toolkit/raw/index.md Read that page and install it.
The prompt includes a link to the full Skill content. You can also view the full content.
Agentic Toolkit is a collection of prompts and tools designed to optimize the development workflow for AI agents (such as Claude Code, Pi, OpenCode, etc.). It primarily addresses issues like insufficient agent planning, unsystematic code review, and lack of automation in CI/CD integration during complex project development.
By providing a series of composable Skills—such as $brainstorm for in-depth planning, $megaplan for managing large projects, $polish-implementation for multi-round PR self-review, and $atk-pr-autofix for automatically fixing CI issues—this toolkit connects scattered agent tasks into a complete, repeatable, and engineered process.
It is well-suited for developers who rely on AI agents for daily coding, project management, and code review, especially for teams or individuals who need to elevate AI-generated output to production-grade quality.
Please note that it is not an out-of-the-box packaged product but rather a reusable "configuration checklist" that requires users to adapt and adjust it based on their chosen agent platform (e.g., Pi requires additional sub-agent configuration) and preferred model.
Key Features
Unlike single-function planning or review Skills, it provides an end-to-end workflow toolchain covering requirement analysis, project planning, code implementation, PR review, and CI monitoring. It particularly excels at improving output quality through multi-round, stateful sub-agent review cycles (the polish series).
Limitations
Not suitable for users seeking an out-of-the-box solution who want full automation without any configuration or understanding; additional sub-agent configuration is required when using it on platforms like Pi.
FAQ
How do I use the polish skills on Pi?
Pi does not have sub-agents by default; you need to install an extension like @tintinweb/subagents first to enable sub-agent functionality.
How do I specify a stronger model during polish-review?
Create an agent named "oracle" in your agent platform and configure it with your preferred high-end model (e.g., Claude Opus).
Installation guide for AI assistants
If your AI coding assistant (Claude Code, Cursor, TRAE etc.) can see this page, send it this message to auto-install:
Visit https://321skill.com/skills/agentic-toolkit/raw/index.md to read the original Skill definition (Markdown format) for agentic-toolkit, and install it according to the instructions.
Raw Markdown URL for AI: /skills/agentic-toolkit/raw/index.md