Skill: Difference between revisions
No edit summary |
|||
| Line 14: | Line 14: | ||
**Preferences | **Preferences | ||
**Procedure | **Procedure | ||
==How does an AI Agent decide to use a Skill== | |||
It decides that automatically, using progressive disclosure. An AI Agent only sees only the names and descriptions of ''active'' skills inside your project session. It adds the skill's instructions to its context when the description matches your request. | |||
==Skill referencing== | ==Skill referencing== | ||
Revision as of 11:00, 15 October 2026
Read more about Beta features.
A Skill item is a structured configuration document that defines what an AI Agent is allowed to do, how it should behave, and which parameters control its generation. Think of it as the "job description" for an AI assistant: instead of shipping a one-size-fits-all chatbot, you add a Skill that scopes the agent's capabilities.
Properties
In the Tygron Platform, skills have three parts:
- Skill name: Used to identify, find and reference the skill.
- Skill description: Used to determine whether to use the skill or not. Should be 1 or 2 sentences max.
- Skill content: The actual skill contents. Subdivided into parts, for example:
- Purpose
- Requirements
- Preferences
- Procedure
How does an AI Agent decide to use a Skill
It decides that automatically, using progressive disclosure. An AI Agent only sees only the names and descriptions of active skills inside your project session. It adds the skill's instructions to its context when the description matches your request.
Skill referencing
A skill can reference an other skill in its content by prepending the name with a '/'. For example, the skill heat-stress-indicator can reference /indicator-excel-sheet to indicate what skill should be used to create a new heat-stress indicator.