The Agent Skills Directory

[PROMPT_INJECTION]: The skill contains deceptive metadata regarding its authorship.
Evidence: The frontmatter author field in SKILL.md is set to "Render", whereas the skill is attributed to "Sarthib7". This discrepancy could mislead users into attributing the skill's instructions to the official vendor.
[PROMPT_INJECTION]: The skill possesses a surface for indirect prompt injection through user-data interpolation.
Ingestion points: The agent is instructed to write and edit render.yaml files based on user requirements provided at runtime.
Capability inventory: Across all referenced documents, the agent is tasked with modifying local filesystem contents (configuration files).
Boundary markers: The instructions do not include delimiters or warnings to treat user-provided configuration values as untrusted content.
Sanitization: No guidance is provided for escaping or validating user inputs before they are interpolated into the YAML structure.
[EXTERNAL_DOWNLOADS]: The skill references external configuration schemas from a well-known service.
Evidence: Provides a link to the official Render JSON schema (https://render.com/schema/render.yaml.json) for validation purposes.

render-blueprints