The Agent Skills Directory

[COMMAND_EXECUTION]: The skill guides the agent to execute a local validation tool via dotnet run --project eng/skill-validator/src/SkillValidator.csproj. This command checks the newly created skill's compliance with repository-specific standards.\n- [INDIRECT_PROMPT_INJECTION]: This skill functions as a template generator that incorporates user-provided inputs into new files.\n
Ingestion points: User-supplied skill name, description, purpose, and workflow steps in SKILL.md.\n
Boundary markers: The generated file uses YAML frontmatter and standard Markdown headers to structure data, but does not use specific adversarial boundary markers for interpolated text.\n
Capability inventory: The workflow involves creating new directories and writing SKILL.md files to the local file system.\n
Sanitization: The skill includes instructions to validate the skill name with regex, enforce character limits on descriptions, and specifically check for (and exclude) secrets, tokens, or XML-like tags.\n- [EXTERNAL_DOWNLOADS]: Provides a link to the agentskills.io domain for documentation regarding the Agent Skills specification.

create-skill