trader-portfolio
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill dynamically installs the
neural-traderpackage from the public npm registry if it is not found locally. It uses the--ignore-scriptsflag to mitigate potential risks associated with pre- or post-install scripts. - [COMMAND_EXECUTION]: Multiple shell commands are executed using
npx neural-traderto optimize allocations, assess risks, and perform portfolio rebalancing. - [INDIRECT_PROMPT_INJECTION]: The skill contains an indirect prompt injection vulnerability surface by fetching historical and current portfolio data and introducing it into the tool execution context.
- Ingestion points: Fetches untrusted data via
mcp__plugin_ruflo-core_ruflo__memory_searchandmcp__plugin_ruflo-core_ruflo__agentdb_pattern-searchinSKILL.md. - Boundary markers: No boundary markers or instructions to disregard embedded commands are present in the text templates.
- Capability inventory: The skill can execute local binary code and shell commands (
npx neural-trader) via the Bash tool. - Sanitization: No active text escaping, filtering, or validation is carried out on the ingested portfolio data before downstream handling.
Audit Metadata