The Agentic Coding Kit

The Agentic Coding Kit is a drop-in configuration pack designed to turn Claude Code and Cursor into more disciplined coding assistants. It provides scoped permissions, task prompts, subagents, review checklists, and command/workflow patterns that keep agent behavior aligned with team conventions.

Inside is a 34-file set you can unzip into your repo and enable by selecting the preset for your stack. It includes CLAUDE.md presets for Python/FastAPI, TypeScript/Node, React/Next.js, Go, and Rust, along with Cursor rules that cover testing, git discipline, security review, refactors, documentation, and style.

You also get Claude Code subagents (code reviewer, test writer, security auditor, refactorer, doc writer), slash commands (/review, /test, /secure, /ship), and hook recipes with ready-to-use scripts (including blocking force pushes, requiring tests before commit, and auto-lint). CI workflow templates, pre-commit config, and ADR templates round out the kit, all delivered in plain Markdown/JSON/shell for flexibility.

What is The Agentic Coding Kit used for?

It’s a configuration pack that adds guardrails and workflows so AI coding tools follow task scope, testing expectations, and team-style conventions.

Which AI coding tools does it support?

It’s intended for Claude Code and Cursor, with presets and rules tailored to their workflows.

What stack presets are included?

There are CLAUDE.md presets for Python/FastAPI, TypeScript/Node, React/Next.js, Go, and Rust. Pick the one that matches your project.

What kinds of guardrails are included?

You’ll find test-first rules, security review patterns, refactor guardrails, documentation/style guidance, and review checklists, plus scoped permissions to limit what the agent can do.

Does it include git hooks or CI templates?

Yes. The kit includes hook recipes (e.g., blocking force pushes, requiring tests before commit, and auto-lint) and CI workflow templates plus pre-commit configuration.

Is it easy to adopt in an existing repo?

Yes—unzip it into your repo, then select the CLAUDE.md preset for your stack and apply the accompanying rules and templates to align the agent from the first prompt.