The Agent Skills Directory

[DATA_EXFILTRATION]: Accesses a sensitive local credential file at ~/.gooseworks/credentials.json to retrieve API keys. While this is used for the service's own authentication, reading credential files is a sensitive operation.
[COMMAND_EXECUTION]: Uses shell commands including curl for network requests and python3 -c for parsing local JSON files. These commands are used for setup and interacting with the backend APIs.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it ingests and processes content from untrusted external PDF URLs.
Ingestion points: Remote PDF URLs provided to the linkup, scrapegraph, and riveter API endpoints in SKILL.md.
Boundary markers: There are no markers or instructions used to delimit the external content from the agent's instructions.
Capability inventory: The skill utilizes shell execution for network operations (curl) and file system access (python3 reading credentials).
Sanitization: No sanitization or validation of the fetched PDF content is performed prior to processing.

pdf-processor