The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill provides a surface for indirect prompt injection by accepting user-supplied strings and embedding them directly into a file's provenance manifest.
Ingestion points: Arguments passed to the /digital-marketing-pro:c2pa-metadata command, specifically --brand, --generator, --prompt, and --reviewer.
Boundary markers: None identified in the skill instructions to delimit untrusted content.
Capability inventory: The skill executes a local Python script scripts/embed-c2pa.py that performs file system write operations to embed metadata.
Sanitization: The instructions do not specify any validation or sanitization for the provided input strings.
[EXTERNAL_DOWNLOADS]: The skill documentation indicates that it automatically installs necessary Python dependencies from the public registry upon execution.
Evidence: Mentions the installation of c2pa-python>=0.5.0 and cryptography via pip.
[COMMAND_EXECUTION]: The skill invokes a local Python script to process assets and embed metadata.
Evidence: Calls scripts/embed-c2pa.py using arguments provided by the user, including file paths for inputs, outputs, and signing credentials.

c2pa-metadata