reading-teacher
Fail
Audited by Gen Agent Trust Hub on Feb 17, 2026
Risk Level: CRITICALCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION] (LOW): The scripts
generate_game.shandgenerate_playground.shuse theevalcommand to process menu selections from theselectcommand. While the current options are hardcoded, the use ofevalon variable content is a poor security practice that could be exploited if the input sources were expanded.\n- [PROMPT_INJECTION] (LOW): The scripts accept a user-provided filename viaread -pand use it to construct a file path without sanitization. This creates a surface for indirect prompt injection via path traversal.\n - Ingestion points:
OUTPUT_FILEvariable in bothscripts/generate_game.shandscripts/generate_playground.sh.\n - Boundary markers: None; the input is directly used in path construction.\n
- Capability inventory: File system write access via
catredirection.\n - Sanitization: Absent; the scripts do not validate the filename or prevent the use of
../sequences to write files outside the intended directories.
Recommendations
- Contains 1 malicious URL(s) - DO NOT USE
Audit Metadata