The Agent Skills Directory

[SAFE]: The skill interacts exclusively with the author's own infrastructure (api.gooseworks.ai) and uses vendor-specific local configuration directory (~/.gooseworks/) for authentication.
[COMMAND_EXECUTION]: Utilizes curl and python3 for legitimate tasks such as fetching API keys from local files and communicating with processing endpoints.
[PROMPT_INJECTION]: As the skill processes external PDF content, it has an inherent surface for indirect prompt injection.
Ingestion points: External document URLs provided as input to the skill during workflow steps.
Boundary markers: No specific delimiters or safety instructions are defined in the workflow prompts to separate data from instructions.
Capability inventory: The skill executes network requests via curl to send data to remote proxy endpoints.
Sanitization: No explicit sanitization or validation of the remote PDF content is performed within the skill instructions, relying on the backend APIs.

pdf-processor