content-topic-plan
Pass
Audited by Gen Agent Trust Hub on Oct 31, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests and processes untrusted data from various external sources to generate content topics.
- Ingestion points:
references/question-sources.mdinstructs the agent to collect data from lead forms, emails, business reviews, competitor reviews, and Google Business Profile performance reports. - Boundary markers: The instructions lack specific delimiters or "ignore embedded instructions" markers when processing the raw customer questions.
- Capability inventory: The skill utilizes tools such as
list_postsandread_post(SKILL.md) and possesses the ability to generate and export data in CSV and ICS formats (references/calendar-export.md). - Sanitization: While
references/question-sources.mdincludes explicit instructions to strip personally identifiable information (PII) such as names, phone numbers, and emails, it does not provide guidance for sanitizing or escaping the content to prevent malicious instructions embedded in the text from influencing the agent's behavior.
Audit Metadata