revisar
Pass
Audited by Gen Agent Trust Hub on Aug 15, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [DYNAMIC_CONTEXT_INJECTION]: The skill utilizes the
!syntax to execute a local Python script (scripts/auditar.py) automatically upon skill loading. While used here for auditing purposes, this mechanism performs silent shell execution and replaces the markdown content with script output without explicit user triggers. - [INDIRECT_PROMPT_INJECTION]: By instructing the agent to 'fix' findings reported by the auditor, the skill creates an indirect injection surface. Maliciously crafted content within the files subjected to audit could theoretically poison the auditor's output to manipulate the agent into performing unauthorized project modifications.
- [COMMAND_EXECUTION]: The frontmatter explicitly defines
allowed-toolsto restrict the agent's shell access to the specific audit script and thetruecommand. This represents a security best practice by adhering to the principle of least privilege.
Audit Metadata