Skills Development

Create, version, document, and validate Ontoly Agent Skills that remain self-contained after installation.

Create a skill under:

skills/<skill-name>/

Required files:

SKILL.md
README.md
examples.md
templates/
reference/

SKILL.md must use YAML frontmatter with name and description, and the name must match the directory.

Public Docs Links

Every skill README.md should link back to the canonical public docs with absolute URLs so installed skills stay self-contained:

https://oss.sarwagya.wtf/ontoly/docs/agent-skills
https://oss.sarwagya.wtf/ontoly/docs/skills-overview
https://oss.sarwagya.wtf/ontoly/docs/mcp
https://oss.sarwagya.wtf/ontoly/docs/capabilities
https://oss.sarwagya.wtf/ontoly/docs/skills-validation

Metadata

Use Ontoly metadata keys:

metadata:
  ontoly.skill.version: "1.0.0-rc.3"
  ontoly.min.version: "1.0.0-rc.3"
  ontoly.capabilities: "ExplainArchitecture, GraphStatistics"
  ontoly.category: "architecture"
  ontoly.enhancement: "LLM Enhancement"
  ontoly.deprecated: "false"

LLM Enhancement is mandatory for official Ontoly skills and every LLM-facing Ontoly workflow. The skill may guide an LLM-capable agent, but all software understanding must still come from Ontoly artifacts, CLI commands, Query Engine results, or MCP capabilities.

Reference Files

The source repository keeps authoring copies under skills/shared/, but each installed skill must be self-contained. Every skill must include and reference:

reference/workflow.md
reference/graph.md
reference/mcp.md
reference/best-practices.md
reference/fallbacks.md
reference/capabilities.md

Do not link from SKILL.md to ../shared/*. That works in the source tree but breaks independent skill installation.

Boundaries

Skills may describe how to invoke Ontoly. Skills must not implement graph understanding, query behavior, parser logic, framework logic, or code generation.

← Skills Best Practices Skills Overview →