The Agent Skills Directory

[COMMAND_EXECUTION]: The batch_design tool accepts a domain-specific "operation script" (DSL) as a string argument. This involves the agent generating and executing a sequence of commands at runtime to perform complex design modifications, creating a dynamic execution surface.
[PROMPT_INJECTION]: The skill possesses a surface for indirect prompt injection by reading data from external .pen files while maintaining powerful write capabilities. Maliciously crafted design files could potentially influence agent behavior when processed through reading tools.
Ingestion points: batch_get, snapshot_layout, get_variables, and get_editor_state (when reading document content).
Boundary markers: Absent. The instructions do not specify the use of delimiters or warnings to ignore instructions embedded within the design data.
Capability inventory: batch_design (scripted design changes), set_variables (metadata modification), replace_all_matching_properties (bulk attribute updates), and open_document (file system interaction).
Sanitization: Absent. There is no mention of validating or escaping content retrieved from design files before it is processed by the agent.
[COMMAND_EXECUTION]: The provided MCP configuration example includes a hardcoded absolute path to a binary in a specific user's directory (/Users/wandl/...). This practice is brittle and poses a risk if the environment is not controlled or if the path points to an untrusted executable.

pencil