tradfi-portfolio-manager
Pass
Audited by Gen Agent Trust Hub on Aug 20, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to execute specific Python scripts located within the local repository (e.g.,
python3 backtests/v3_allocate_today.py) to determine portfolio state and allocations. - [COMMAND_EXECUTION]: References a hardcoded absolute path to a Python virtual environment (
/Users/engineer/.venv/bin/python3), which discloses specific user profile information and assumes a fixed directory structure on the host machine. - [PROMPT_INJECTION]: The skill presents an indirect prompt injection surface due to its requirement to fetch external URLs and process verbatim quotes for citations.
- Ingestion points: External URLs accessed via web tools to fulfill the "Citation rule" requirement.
- Boundary markers: None. The skill requires verbatim quotes without specifying delimiters to isolate untrusted external content from the agent's instructions.
- Capability inventory: The agent has the ability to execute local Python scripts, read local repository files (
GOAL.md,strategy/*.md), and perform network operations to fetch URLs. - Sanitization: No sanitization, validation, or filtering of the external content is mandated before it is processed by the agent.
Audit Metadata