deepsearch
Pass
Audited by Gen Agent Trust Hub on Feb 16, 2026
Risk Level: LOWPROMPT_INJECTIONNO_CODE
Full Analysis
- PROMPT_INJECTION (LOW): The skill creates an attack surface for indirect prompt injection by processing external codebase content. While the capabilities are limited to reading and synthesizing information, there are no instructions to sanitize or ignore embedded commands in searched files. 1. Ingestion points: File content read during the broad search and deep dive phases in SKILL.md. 2. Boundary markers: Absent; no delimiters or instructions to ignore embedded prompts are provided. 3. Capability inventory: Reading files and generating summaries. 4. Sanitization: Absent.
Audit Metadata