explore-code
Warn
Audited by Gen Agent Trust Hub on Apr 5, 2026
Risk Level: MEDIUMCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The script
scripts/write_outputs.pyusesimportlib.utilto dynamically load and execute a module from a path computed at runtime (../../../shared/scripts/write_explore_bundle.py). This dynamic execution of code from outside the skill's package can be used to execute logic that is not visible during a static review of the skill's content. - [PROMPT_INJECTION]: The skill is susceptible to Indirect Prompt Injection due to its interaction with external research repositories.
- Ingestion points: The skill ingests source code and markdown from research repositories mentioned in
SKILL.md. - Boundary markers: There are no defined delimiters or instructions to ignore embedded prompts in
SKILL.mdorreferences/explore-policy.md. - Capability inventory: The skill can modify branches, write output files to
explore_outputs/, and hand off execution to training tools likerun-train. - Sanitization: No sanitization or validation of the external repository content is documented.
Audit Metadata