software-workflow-automation
Pass
Audited by Gen Agent Trust Hub on Oct 6, 2026
Risk Level: SAFECOMMAND_EXECUTIONDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The script
scripts/replay_dlq.pyutilizessubprocess.runto execute system commands. Although the default configuration uses the benignechocommand, the script is explicitly designed as a scaffold to be modified by users to invoke other system tools liketemporal,aws, orcurlfor replaying messages. - [DYNAMIC_EXECUTION]: The
scripts/replay_dlq.pyscript assembles command strings at runtime by interpolating data from an external JSON file into a command template, which is then executed by the system shell. - [INDIRECT_PROMPT_INJECTION]: The
scripts/replay_dlq.pyutility presents a surface for indirect injection as it processes data from external JSON files to drive system actions. - Ingestion points: The script reads message objects from a JSON file provided via the
input_fileCLI argument. - Boundary markers: There are no boundary markers or explicit instructions provided to the agent or script to ignore malicious instructions embedded within the JSON payload.
- Capability inventory: The script has the capability to execute arbitrary commands on the host system via
subprocess.run(identified inscripts/replay_dlq.py). - Sanitization: The script uses
json.dumps()to format the payload before insertion into the command template, which provides basic structural escaping but does not sanitize the logical content of the message against command-specific injection vectors.
Audit Metadata