review-codebase
Warn
Audited by Gen Agent Trust Hub on Mar 18, 2026
Risk Level: MEDIUMCOMMAND_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill requires the
Bashtool for inventorying files. A shell environment enables the execution of arbitrary commands on the host system beyond the intended scope. - [DATA_EXFILTRATION]: The skill is instructed to audit for hardcoded secrets, API keys, and credentials. When combined with the
WebFetchtool, this creates a significant exfiltration surface for sensitive data found during the review. - [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection through the analysis of untrusted files in the
target_path. 1. Ingestion points: Files read usingRead,Grep, andGlobwithin the user-specified directory. 2. Boundary markers: None; the skill lacks instructions to distinguish between the agent's core logic and instructions potentially embedded in the audited code. 3. Capability inventory:Read,Grep,Glob,Bash, andWebFetch. 4. Sanitization: None; data ingested from the codebase is not validated or sanitized before being processed by the agent.
Audit Metadata