asd-ste100-hebrew
Pass
Audited by Gen Agent Trust Hub on Oct 9, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes arbitrary Hebrew text provided by users or agents to perform rewrites, which presents a surface for indirect prompt injection.
- Ingestion points: User-supplied text is ingested and processed as defined in the
SKILL.mdinstructions. - Boundary markers: The instructions constrain the output to the rewritten text but do not implement explicit input sanitization or prompt escaping.
- Capability inventory: The skill utilizes a local Python script (
scripts/ste-lint-he.py) to perform structural analysis of the text. - Sanitization: No sanitization or filtering is performed on the input text before it is evaluated by the agent.
- [DYNAMIC_EXECUTION]: The unit test suite (
tests/test_ste_lint_he.py) usesimportlib.utilto dynamically load and execute the linter script (scripts/ste-lint-he.py) for testing purposes. - [COMMAND_EXECUTION]: The test suite (
tests/test_ste_lint_he.py) uses thesubprocess.runfunction to execute the linter script as a separate process using the current Python interpreter (sys.executable).
Audit Metadata