skill-publish-scheduler
Pass
Audited by Gen Agent Trust Hub on Oct 16, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests scheduling data from CSV and JSON files, including fields like 'content' and 'title' which are subsequently passed to other publishing skills. This creates an attack surface for indirect prompt injection if the downstream tools do not properly sanitize the inputs.
- Ingestion points: Data is ingested through the
importandaddcommands inscripts/publish_queue.pywhich read from external files or user input. - Boundary markers: There are no specific delimiters or instructions provided to the agent to treat the imported content as untrusted data.
- Capability inventory: The skill performs file read and write operations for managing its local queue in
scripts/publish_queue.pybut does not perform network operations or direct command execution. - Sanitization: The script performs date format validation but does not sanitize, filter, or escape the text content, titles, or notes fields.
Audit Metadata