prompt-injection-risk-linter
Pass
Audited by Gen Agent Trust Hub on May 12, 2026
Risk Level: SAFE
Full Analysis
- [PROMPT_INJECTION]: The skill contains phrases such as 'ignore previous instructions' and 'system prompt' within its documentation and internal regex patterns. These are used solely for the tool's intended purpose of detecting prompt injection markers and do not constitute instructions for the agent to bypass its own safety guidelines.
- [DATA_EXFILTRATION]: The script reads local input files and writes a JSON report to a specified local path. It does not contain any network communication, API requests, or logic for accessing sensitive system credentials.
- [REMOTE_CODE_EXECUTION]: The implementation is restricted to the Python standard library (argparse, json, re). It does not use dynamic code execution functions like 'eval()' or 'exec()', nor does it utilize 'subprocess' modules to run external code.
- [COMMAND_EXECUTION]: The script does not invoke external shell commands or interact with the operating system beyond basic file I/O.
- [DATA_EXPOSURE]: The script processes files provided via command-line arguments. While it handles potentially untrusted content, its operations are limited to pattern matching and generating a report, posing no risk of exposing sensitive system data or environment variables.
Audit Metadata