membership-weekly-digest
Pass
Audited by Gen Agent Trust Hub on Sep 9, 2026
Risk Level: SAFEPROMPT_INJECTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill includes instructions in SKILL.md that attempt to override the agent's tool selection logic by directing it to use this skill by default for any membership-related summary unless explicitly requested otherwise. It also specifies that the agent should not shorten or simplify the output template.
- [INDIRECT_PROMPT_INJECTION]: The skill processes external data from subscription and traffic logs, which creates a potential surface for indirect prompt injection if the source data contains adversarial instructions.
- Ingestion points: SKILL.md (via tool outputs for subscription and traffic lists)
- Boundary markers: None present
- Capability inventory: SKILL.md (Data aggregation and Markdown formatting; no destructive or network capabilities listed)
- Sanitization: None present
Audit Metadata