Mikros: Skill Library for AI Programming Agents and Megálos Workflow Framework

Mikros is an innovative agent-skills library designed for Megálos workflows. It uses Markdown instead of code to enable AI programming agents to consume and understand workflow definitions, supporting full-process automation of creation, testing, validation, and deployment. The naming philosophy reflects its core idea: building grand workflow systems through small, refined skill units (Mikros means "tiny" in Greek, Megálos means "grand").

With the rapid development of AI coding agents, organizing and transferring workflow knowledge has become a key challenge. Traditional code libraries require direct calls by human developers, while AI agents need a more declarative, easy-to-understand knowledge representation. Mikros was created to address this problem.

Mikros is an agent-skills library dedicated to creating Megálos workflows. Its core features include:

• Markdown-first: All skills exist as Markdown documents instead of executable code.
• AI consumption-oriented: Content is consumed by AI programming agents rather than directly called by humans.
• Dynamic expansion: New skills are added as Markdown files to the skills/ directory.

Skill Organization

Mikros uses a hierarchical skill organization strategy. Each skill is an independent Markdown document containing: purpose description, usage scenarios and examples, input/output specifications, and relevant context information. This allows AI agents to select and apply appropriate skills via semantic understanding instead of hard-coded API calls.

Adapter Mechanism

The project defines multiple adapter specifications through docs/ADAPTERS.md, explaining how different types of AI agents consume the skill library. This design enables:

• Different AI agent architectures to use the same skill library.
• Decoupling of skill definitions and consumption methods.
• Community contributions of new adapter implementations.

Automated Workflow Creation

Mikros enables AI agents to:

1. Understand complex workflow requirements.
2. Retrieve relevant skills from the library.
3. Combine multiple skills to form a complete workflow.
4. Automatically generate, test, and deploy workflow code.

Knowledge Reuse & Sharing

By encoding workflow knowledge into Markdown skill documents, teams can:

• Precipitate best practices.
• Reuse workflow patterns across projects.
• Reduce learning costs for new members.
• Implement versioned management of workflow knowledge.

Mikros represents an emerging shift in AI development paradigm: from "humans write code for machines to execute" to "humans define knowledge for AI to understand and apply". This shift has the following significance:

1. Higher-level abstraction: Developers focus on "what to do" rather than "how to do it".
2. Natural language interface: Use Markdown and natural language to describe workflows.
3. Adaptive execution: AI agents dynamically select execution strategies based on context.
4. Continuous learning: The skill library can be continuously expanded, and AI agents automatically gain new capabilities.

Mikros is open-sourced under the MIT license, reflecting the project team's commitment to open collaboration. This openness helps:

• Build a community ecosystem around Megálos workflows.
• Promote standardization of skill documents.
• Drive the popularization of AI programming agent technology.

The Mikros project demonstrates a forward-looking AI development paradigm. By encoding workflow knowledge into AI-consumable Markdown documents, it opens up new possibilities for automated software engineering. As the capabilities of AI programming agents continue to improve, skill library frameworks like Mikros will play an increasingly important role in future software development.