gnuplot
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes data from the current conversation context to generate plots. This data is untrusted and could contain characters meant to escape the gnuplot script's data blocks (e.g., terminating a heredoc prematurely).
- Ingestion points: Data derived from the interaction context (SKILL.md).
- Boundary markers: Uses heredocs ($DATA << EOD) which can be bypassed if the input contains the delimiter.
- Capability inventory: Executes shell commands via Bash tool and runs gnuplot on generated files.
- Sanitization: The skill does not define specific sanitization or escaping for the data content before interpolation.
- [DYNAMIC_EXECUTION]: The skill programmatically generates gnuplot scripts and saves them to temporary files before execution.
- [COMMAND_EXECUTION]: Executes shell commands including gnuplot and emacsclient --eval to interact with the local environment for styling and plotting.
Audit Metadata