jinko-output-set
Pass
Audited by Gen Agent Trust Hub on Oct 23, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to process user-supplied mathematical formulas and constraints (e.g.,
age >= 18,auc(Drug)). This represents a data ingestion surface where external instructions could potentially be embedded. - Ingestion points: Formula and constraint strings provided via user prompts or external JSON files (
assets/advanced_output_set_example.json). - Boundary markers: The instructions recommend explicit validation calls (
client.validate_scoring_formula) and check diagnostics to mitigate syntax errors or malformed input. - Capability inventory: The skill uses the Jinkō SDK to communicate with a remote modeling platform and includes CLI tools for creating and editing cloud-based resources.
- Sanitization: The skill relies on platform-side validation and syntax checks provided by the SDK to ensure inputs conform to the Jinkō formula grammar.
- [COMMAND_EXECUTION]: The documentation provides multiple examples of executing shell commands via Python's module interface (e.g.,
python -m jinko.cli.create_simple_output_set). These commands are standard for a developer-oriented SDK and allow users to interact with the platform from the command line. - [EXTERNAL_DOWNLOADS]: The skill references the installation of the
jinko-sdkand additional skills from the vendor's repository (novainsilico/jinko-skills). These references point to official resources owned by the skill's author and are used for legitimate setup and dependency management.
Audit Metadata