refine
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted content from an external file, which introduces an indirect prompt injection vulnerability surface if malicious instructions are embedded within the todo items.
- Ingestion points: Reads items from the external data file
brain/todos.mdas described inSKILL.md. - Boundary markers: Absent; there are no delimiters or explicit instructions telling the agent to ignore any potential system commands or instructions inside
brain/todos.md. - Capability inventory: Possesses the capability to modify files via the
Edittool and communicate with the user viaAskUserQuestion. - Sanitization: Absent; the content is parsed and rewritten directly into the workspace without validation or sanitization filters.
Audit Metadata