The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to process and tune OpenAPI/AsyncAPI specification files, which serve as an untrusted data ingestion surface.
Ingestion points: The skill ingests OpenAPI and AsyncAPI specification files referenced in fern/generators.yml and throughout the SKILL.md working rules.
Boundary markers: The instructions do not define specific delimiters or warnings for the agent to ignore potentially malicious instructions embedded in the natural language fields of the specifications, such as description, summary, or example values.
Capability inventory: The skill utilizes the fern CLI tool, which provides capabilities for network operations (fern api update), file writes (fern write-overrides, fern write-definition), and local project initialization (fern init).
Sanitization: There is no evidence of sanitization or filtering for external content within the specification files being processed, allowing for potential manipulation of the agent's behavior during the editing or generation workflow.

fern-spec