Skip to main content
Skills are folders of instructions, scripts, and resources that agents can discover and use to perform tasks more accurately and efficiently. They provide procedural knowledge and context that agents can load on demand.

Why Skills?

Agents are increasingly capable, but often lack the context needed to do real work reliably. Skills solve this by giving agents access to:
  • Domain expertise: Specialized knowledge packaged into reusable instructions
  • New capabilities: Extended functionality like creating presentations, building MCP servers, or analyzing datasets
  • Repeatable workflows: Consistent, auditable multi-step task execution
  • Organizational knowledge: Team and company-specific context in portable, version-controlled packages

How Skills Work in Zencoder

Zencoder automatically discovers and loads skills from these locations: The agent decides which skill to use based on the skill’s description and the current task context. Skills are available in both Zenflow and Zencoder IDE Agents.
Skills are automatically selected by the agent based on task context. Manual skill selection is not currently supported.
The legacy .zencoder/skills/ path is deprecated but still supported. We recommend migrating to .agents/skills/ to stay aligned with the current standard.

Skill Structure

A skill is a folder containing a SKILL.md file with metadata and instructions:

Creating a Skill

Create a SKILL.md file in your skills folder:

SKILL.md Frontmatter Reference

The description field is critical — write it as a clear trigger condition. For example: “Use this skill when the user asks to create or modify API endpoints” is more effective than “API helper skill.”

Example Skills

A skill that enforces your team’s review standards on every code review request.
A skill that generates migration files following your ORM’s conventions.
A skill that creates REST endpoints following your team’s patterns.

Skills vs AI Agents

Tips for Effective Skills

The agent matches skills to tasks based on the description. Be explicit: “Use when the user asks to generate unit tests for React components” works better than “Testing helper.”
If a skill only applies to your API layer, add paths: [src/api/**] so it doesn’t activate for frontend tasks.
Show the agent what good output looks like. A “before and after” example in the instructions is worth more than a paragraph of explanation.
Keep skills in your repo (.agents/skills/). They evolve with your codebase and benefit from code review like any other file.
Zenflow Settings showing Prompts & Skills page with saved prompts and skills sections

Built-In Orchestration Skills

Zenflow includes built-in skills for multi-agent orchestration:

Skills vs Saved Prompts

Learn More

Skills follow an open standard developed by Anthropic and adopted by leading AI development tools. For the complete specification and examples, see the Agent Skills documentation.