Agent Operating Kit: A Collection of Documents and Process Templates for Collaborating with AI Programming Assistants

Agent Operating Kit (AOK) is a collection of document templates and process specifications designed to help developers collaborate efficiently with AI programming assistants. It aims to address common operational issues in AI-assisted development (such as inconsistent documentation, chaotic workflows, and excessive context size). Its core value lies in reducing the context consumption of AI assistants, enhancing collaboration consistency, and protecting existing project assets through streamlined document structures, standardized workflows, and component-based selection.

With the popularity of AI programming assistants like Claude, Cursor, and Codex, teams face the following operational issues:

1. AI does not know where to start implementing planning documents
2. Tasks lack consistent processes
3. AI-generated documents deviate from code, leading to inconsistency between documents
4. AGENTS.md or planning documents are too large, causing small tasks to consume a lot of context
5. Concerns about losing important details when introducing AI rules to existing projects

AOK was created to address these issues. It is not a larger prompt, but a collection of reusable documents and workflow templates that solidify best practices for AI collaboration.

Core Concepts

1. Keep Always-Read Documents Concise: The root directory's AGENTS.md is limited to 50-100 lines, with detailed content moved to independent documents for on-demand reading
2. Component-Based Selection: Choose components as needed to avoid unnecessary complexity
3. Protect Existing Assets: Read existing documents first when applying; modifications require user approval
4. Reduce Default Context Consumption: The success indicator is a reduction in default context for daily tasks

Core Components

• Root Directory AGENTS.md: Always read by AI, including project overview, common commands, verification processes, and document routing
• DESIGN.md: UI/frontend design benchmarks (colors, fonts, layout, etc.)
• Document Map and Governance: .agents/INDEX.md indexes workflows, defines authoritative sources and change approval rules
• Workflow Documents: .agents/workflows/*.md (standardized steps for feature development, bug fixes, etc., based on the Superpowers framework)
• Decision Records and Status Tracking: ADR, current status, version history, work logs

New Project Application Process

1. User requests to apply AOK
2. AI reads the agentOperatingKit.md in the corresponding language
3. Displays a list of optional components
4. User selects components
5. AI reads the selected templates
6. Generates a plan and implements it after user approval

Existing Project Migration Principles

1. Read existing documents first
2. Identify information distribution
3. Propose a migration plan (explain files to modify, move, or delete)
4. Obtain user approval
5. Progressive migration (apply components in phases)

Scenario 1: Personal Project Initialization

Xiao Ming uses the AOK lite version: root directory AGENTS.md (project overview, common commands, test process), no workflows, reducing context consumption

Scenario 2: Team Project Standardization

A 5-person team uses the full version: unified AGENTS.md, standardized workflows, DESIGN.md, ADR documents, ensuring consistent task execution

Scenario 3: AI Transformation of Existing Project

2-year project progressive migration:

1. Phase 1: Create a concise AGENTS.md
2. Phase 2: Introduce new feature workflows
3. Phase 3: Integrate specifications into DESIGN.md The process follows the 'read first → analyze → migrate' principle

AGENTS.md Writing Recommendations

1. Keep concise, only include information AI always needs
2. Clear document routing rules
3. Include task verification steps
4. Include version control

Workflow Usage Recommendations

1. Create workflows only for repetitive tasks
2. Keep composable
3. Set key checkpoints
4. Clear output requirements

Document Governance Recommendations

1. Define authoritative source documents
2. Establish AI document change approval process
3. Regularly review document consistency
4. Use ADR to record important decisions

AOK's core values:

1. Reduce AI context consumption
2. Enhance collaboration consistency
3. Protect existing project assets
4. Flexible and scalable (component-based selection)

Outlook: As AI programming assistants become more popular, efficient collaboration will become an important skill, and AOK provides a practical methodology. It is recommended to start with a concise AGENTS.md and gradually introduce components. Remember the core principle: a successful application should reduce default context consumption.

Project resources: Bilingual maintenance (English/Korean), core documents include agentOperatingKit.md, application guide, component selection guide, etc.