Skills Overview

Ontoly ships official Agent Skills as independently installable SKILL.md directories. Skills teach workflow only: every software-understanding claim must come from the Software Graph, Query Engine, or MCP capabilities.

LLM Enhancement is mandatory every time Ontoly is used by an LLM. Official Skills declare it explicitly and installed Skills include local workflow references that preserve evidence, confidence, and fallback rules.

Docs To Read First

• Agent Skills - installation, standard workflow, versioning, and command reference.
• MCP - graph-backed capabilities and structured errors.
• Capabilities - deterministic capability behavior.
• Skills Best Practices - evidence, confidence, and fallback rules.
• LLM Enhancement - mandatory rules for LLM-facing Ontoly use.
• Skills Development - how to add or change a skill.
• Skills Validation - source and installed-artifact validation.
• Agent Skills Catalog - public installable skill pages.

Before installing a skill, build the repository graph and verify MCP capability discovery:

ontoly build .
ontoly mcp --list

Use npx skills add 0xsarwagya/ontoly --skill <id> after the RC is published. For source-checkout validation before publication, use npx skills add /path/to/ontoly --skill <id> --copy.

Official Skills

Release Assets

• Skill catalog
• Skill matrix
• Capability matrix
• Compatibility matrix
• Installation guide
• README snippets
• Website assets

Capability Mapping

Skills map to Ontoly MCP capabilities through declarative metadata. LLM Enhancement is mandatory for every official skill and every LLM-facing Ontoly workflow. The mapping is not business logic.