algo-risk-test
Pass
Audited by Gen Agent Trust Hub on Sep 17, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes strategy files to extract parameters (e.g., SYMBOL, RISK) which are then interpolated into a generated verification script. This creates a surface where malicious content in a strategy file could influence the generated script.
- Ingestion points: The skill reads strategy files provided via the
[strategy-file]argument inSKILL.md. - Boundary markers: Absent. The extraction logic lacks delimiters or warnings to ignore instructions embedded within the strategy data.
- Capability inventory: The skill uses
Read,Write, andBashtools. The generatedrisk_test.pyscript performs file system operations and local network requests to the OpenAlgo sandbox. - Sanitization: None detected. Extracted values are directly inserted into the Python script template.
- [DYNAMIC_EXECUTION]: The generated
risk_test.pyscript template includes a loop that traverses parent directories to find a specific path (.claude/skills/algo-expert/rules/assets/core) and dynamically modifiessys.pathto import modules from that location. - Evidence: The script uses
sys.path.insert(0, str(c.parent))after dynamically verifying the existence of the directory structure.
Audit Metadata