vuln-scan
Pass
Audited by Gen Agent Trust Hub on Oct 29, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted source code and
THREAT_MODEL.mdfiles, interpolating their content directly into instructions for subagents. - Ingestion points: Target source files and
THREAT_MODEL.md(specifically the "Entry points & trust boundaries" and "Threats" sections) inSKILL.mdStep 1 and Step 2. - Boundary markers: The skill uses structural headings (e.g.,
TARGET:,TASK:) and XML-style tags (<finding>) to delimit inputs and outputs, but does not provide specific instructions to the subagents to ignore or treat embedded instructions in the target code as data rather than instructions. - Capability inventory: The skill and its subagents have access to
Read,Glob,Grep,Write,Task, and restrictedBashcommands (rg, grep, ls, wc, head, file). - Sanitization: There is no evidence of sanitization, escaping, or validation of the content read from the target directory before it is included in the subagent prompts in Step 2.
Audit Metadata