The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from an external source (Figma) which could theoretically contain malicious instructions designed to influence the agent.
Ingestion points: Untrusted content including user comments, file names, and component descriptions is fetched from the Figma API in figma_skill.py.
Boundary markers: The skill outputs all data as structured JSON using json.dumps, which helps the agent distinguish between tool output and instructions.
Capability inventory: The skill's environment includes Bash and Read tools, providing a capability surface that could be targeted by successful injection.
Sanitization: Content retrieved from the API is not sanitized for potential prompt injection patterns before being returned to the agent.
[EXTERNAL_DOWNLOADS]: The skill relies on the requests Python library to communicate with the Figma API.
Evidence: The script checks for the presence of the requests package and provides a standard installation command (pip install requests) if it is missing.

figma-skill