The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes a local utility script, scripts/studio-run.mjs, to handle project initialization, stage advancement, and JSON data validation. These operations are restricted to the local project directory and are consistent with the skill's functional purpose.
Evidence: SKILL.md contains multiple instructions to execute node <skill-dir>/scripts/studio-run.mjs for tasks like init, advance, and select.
[INDIRECT_PROMPT_INJECTION]: The skill processes potentially untrusted external data in the form of product briefs and design references, which are then used to influence the agent's creative output.
Ingestion points: Input brief provided during init (stored in brief.md) and external references captured in references.json.
Boundary markers: The skill instructions do not explicitly define delimiters for untrusted content, though it uses structured JSON schemas for data storage.
Capability inventory: The skill is capable of reading and writing multiple JSON manifest files within the .styleseed project directory but does not perform network operations or high-privilege system tasks.
Sanitization: The slugify function in studio-run.mjs is used to sanitize directory names generated from user input, reducing the risk of path traversal.

ss-studio