Practical AI agent workflow kits

Reusable project systems for agent-operated work.

Agent Kits Lab creates copy-ready templates, rules, commands, agents, hooks, and operating guides for developers and small teams building with AI coding agents.

Buy the $9 Starter Kit Get the free GitHub template 
claude-code-project-starter-kit/
├── CLAUDE.md
├── .mcp.json.example
└── .claude/
    ├── settings.json
    ├── rules/
    ├── commands/
    ├── agents/
    ├── hooks/
    └── skills/

What we build

Small, auditable workflow assets that help teams move from ad-hoc prompting to repeatable agent-assisted execution.

Project starter kits

Opinionated file structures for Claude Code, Codex, Hermes-style orchestration, and multi-agent engineering workflows.

Rules and review systems

Reusable coding rules, testing expectations, security boundaries, review commands, and release checklists.

Agent operation guides

Plain-language guides for using AI agents responsibly in software projects without overpromising outcomes.

First kit

Claude Code Project Starter Kit

A practical template for making Claude Code behave more like a structured project teammate.

Buy on Ko-fi — $9 Free version

Included in the Starter package

  • Full CLAUDE.md operating guide template
  • Modular rules for code style, testing, API, security, and docs
  • Custom commands for review, issue fixing, planning, docs, and release checks
  • Specialized agents for code review, security audit, test design, docs, and release QA
  • Hook examples for safer bash validation and post-edit reminders
  • MCP config example and structure diagram

The paid ZIP is now available on Ko-fi for $9. You can also start with the free GitHub template and upgrade when you want the complete starter package.

Built for practical developer workflows

Focused templates and operating guides for teams that want clearer project context, safer reviews, and more repeatable AI-assisted development.

Project-ready assets

Copy-ready files, examples, and checklists that can be adapted to real repositories without starting from a blank page.

Human approval where it matters

Public releases, account actions, payments, credentials, and customer commitments stay review-first.

Clear expectations

Templates improve workflow consistency, but every team should still review outputs, run tests, and adapt rules to its own stack.