saas-marketing-suite
Pass
Audited by Gen Agent Trust Hub on Oct 12, 2026
Risk Level: SAFEPROMPT_INJECTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill instructions state that the 'Rules for agents reading this file' section in the external 'brand-context.md' file overrides the agent's own defaults. This directive instructs the agent to prioritize external, potentially untrusted input over its own pre-defined operational boundaries.
- [INDIRECT_PROMPT_INJECTION]: The skill possesses a documented attack surface for indirect prompt injection via local file ingestion.
- Ingestion points: The skill is designed to search for and read 'brand-context.md' from the project root, './.claude/', or './docs/'.
- Boundary markers: None are present; the skill is explicitly told to treat the file content as 'binding' for the duration of the run.
- Capability inventory: The skill routes marketing tasks across 17 specialties involving 90 different agents, meaning any instructions injected into the context file are propagated to a wide range of specialized agents.
- Sanitization: The skill instructions recommend reading the file 'in full' and treating it as binding without any verification or sanitization of the content before it influences the agent's behavior.
Audit Metadata