OPDEEP AI: AI-Native Project Management for Coding Agents

Operational Deep Engineering Protocol for AI

Stop managing AI agents like human teams.

OPDEEP AI is an AI-native project-management and context-engineering framework for coding agents, low-context AI agents, agentic software development, and AI-native SDLC workflows. It turns messy repositories into AI-maintainable modular systems with business module file manifests, task cards, latest-pointer handoffs, local tool routing, verification gates, and reusable experience memory.

Humans struggle to manage hundreds of small project files. AI agents can search, index, update, and assemble them. OPDEEP AI turns that AI advantage into the project architecture.

Why OPDEEP AI

Most project-management systems were designed for humans: meetings, long specs, broad epics, tribal memory, and large documents. That is not how coding agents work best.

AI agents need:

• small indexed files instead of giant documents;
• latest pointers instead of reading full history every time;
• business module manifests before editing code;
• task cards small enough for low-context AI to finish safely;
• explicit read/write boundaries;
• local tools before context-heavy reading;
• verification gates instead of "looks good";
• experience memory that survives the next thread.

OPDEEP AI is the Jens Method packaged as a reusable skill and open framework.

The Three Gates

Phase 1: Analysis Only

Analyze requirements, search mature solutions online, compare technical routes, and write manuals. No implementation code, no scaffolding, no package installation, no task cards.

Phase 2: Environment And Skills

Guide the user to install required software, CLIs, dependencies, plugins, connectors, and AI skills. Record setup and verification results.

Phase 3: AI-Native Project Management

Create the AI-maintenance system:

• source boundary reports;
• business module file manifests;
• project indexes and context packs;
• task board and entity task cards;
• latest-pointer handoffs;
• verification gates and quality checklists;
• role/skill boundaries;
• migration transition rules;
• experience memory.

Core Blocks

launcher -> phase index -> latest setup docs -> file manifests -> task board
-> task card -> verification report -> handoff snapshot -> experience memory

Repository Contents

SKILL.md                       # Codex skill entry
agents/openai.yaml             # Skill UI metadata
references/                    # OPDEEP AI playbooks and templates
scripts/audit_storage_route.py # Checks launcher/latest/history structure
scripts/audit_task_pack.py     # Checks task board and task card safety
llms.txt                       # AI-search and LLM crawler summary
docs/LAUNCH_COPY.md            # Public launch copy and positioning

Install As A Codex Skill

Clone this repository, then copy or symlink it into your Codex skills folder:

git clone https://github.com/JensFeng/opdeep-ai.git
mkdir -p ~/.codex/skills
ln -s "$PWD/opdeep-ai" ~/.codex/skills/opdeep-ai

Then invoke:

Use $opdeep-ai to design an AI-native modular project-management workflow for my next project.

Use Without Codex

You can still use OPDEEP AI as a framework:

1. Read SKILL.md.
2. Use references/task-pack-template.md to create an ai-maintenance/ folder.
3. Use references/business-module-manifests.md before business-code work.
4. Run the audit scripts before handing tasks to a low-context AI.

python3 scripts/audit_storage_route.py <ai-maintenance-dir>
python3 scripts/audit_task_pack.py <task-pack-dir>

What Makes It Different

OPDEEP AI does not ask AI agents to behave like human project managers. It redesigns project management around the things AI agents are good at:

• retrieval;
• indexing;
• bounded context;
• modular composition;
• rule-following;
• repetitive local validation;
• structured handoff.

Keywords For AI Search

AI-native project management, context engineering, coding agents, AI coding agents, low-context AI agents, agentic software development, AI-native SDLC, agent workflow, agent orchestration, task cards, file manifests, business module manifests, handoff protocol, latest pointer, local tool routing, verification gates, experience memory, modular AI maintenance, OPDEEP AI, Jens Method.

中文简介

OPDEEP AI 是一套面向 AI 编程助手的原生项目管理协议。它的核心观点是：

不要再用人的项目管理方式管理 AI。

人不擅长管理大量细碎文件，但 AI 擅长检索、索引、更新和拼装细文件。OPDEEP AI 把这个特点变成项目管理结构：启动器、最新指针、文件清单、业务模块 manifest、任务卡、验证门、交接快照、本地经验库。

License

MIT