The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted metadata from the Pascal scene environment, which introduces an indirect prompt injection surface.\n
Ingestion points: Untrusted data is ingested via get_scene and get_node tools (SKILL.md, references/evidence-boundaries.md).\n
Boundary markers: The skill contains explicit instructions to treat all imported metadata as data, stating it cannot authorize project changes or override skill instructions (SKILL.md).\n
Capability inventory: The agent possesses capabilities to modify the project using place_item and to persist changes via validate_scene (SKILL.md, references/report-template.md).\n
Sanitization: No programmatic sanitization is specified; the skill relies on the agent's adherence to instructional boundaries.\n- [EXTERNAL_DOWNLOADS]: The skill's setup instructions direct the user to install a global NPM package from the vendor's namespace.\n
Evidence: references/setup.md includes the command npm install --global @pascal-app/cli@beta.\n
Context: The package is an official resource from the vendor (@pascal-app) and is used for establishing the connection to the service.

furniture-fit