indicator-design
Pass
Audited by Gen Agent Trust Hub on Sep 18, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [SAFE]: The skill consists of markdown instructions and code templates intended for the AI agent to follow. It does not include any executable scripts, remote resource downloads, or access to sensitive system resources.\n- [INDIRECT_PROMPT_INJECTION]: The skill possesses a surface for indirect prompt injection as it ingests user-provided trading hypotheses to generate indicator code.\n
- Ingestion points: Untrusted user input is captured during the 'Extract the hypothesis' and 'Operationalize' steps in SKILL.md.\n
- Boundary markers: No explicit boundary markers or 'ignore' instructions are used to delimit user input from the generated code.\n
- Capability inventory: The skill defines capabilities for generating vectorized pandas code and performing correlation analysis; it lacks network access, file system write permissions, or subprocess execution capabilities within SKILL.md.\n
- Sanitization: The skill relies on the AI agent's internal reasoning to translate natural language into code, providing a layer of mediation between the user input and the resulting output.\n- [DYNAMIC_EXECUTION]: The skill instructs the agent to generate and present Python code snippets based on user input.\n
- Evidence: SKILL.md contains Python templates for a pandas-based indicator function and sanity-check print statements that the agent is expected to populate at runtime.
Audit Metadata