site-architecture
Pass
Audited by Gen Agent Trust Hub on Oct 2, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill instructions in
SKILL.mddirect the agent to read and incorporate content from external files (.agents/product-marketing-context.mdor.claude/product-marketing-context.md) if they exist. This creates a vulnerability surface where a user or an attacker could place malicious instructions inside these marketing context files to influence the agent's behavior during the planning process. - Ingestion points:
SKILL.md(instructions to read external marketing context files). - Boundary markers: Absent; there are no specific delimiting instructions or warnings to the agent to treat the content of the marketing context files strictly as non-executable data.
- Capability inventory: The skill instructions are limited to generating site structure documentation (ASCII trees, Mermaid diagrams, and URL maps) and do not invoke dangerous tools like arbitrary shell command execution or network exfiltration.
- Sanitization: Absent; the skill does not suggest any validation or filtering of the content read from external context files.
Audit Metadata