longbridge-value-investing
Pass
Audited by Gen Agent Trust Hub on Oct 12, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill uses the
longbridgeCLI to fetch financial reports, snapshots, and ownership data. These commands are structured, use specific subcommands (e.g.,financial-report,calc-index), and do not incorporate unvalidated user input into shell arguments, minimizing command injection risks. - [INDIRECT_PROMPT_INJECTION]: The skill processes data from external sources (Longbridge CLI and WebSearch).
- Ingestion points: Data enters the context via CLI outputs and WebSearch results used for industry outlooks and qualitative signals.
- Boundary markers: The instructions mandate a structured output template and a 'Data Source Appendix' to separate source data from the analysis.
- Capability inventory: The agent has the ability to execute CLI tools and perform network searches via the platform's tools.
- Sanitization: The skill implements a 'Reconciliation Gate' that performs mathematical consistency checks (e.g., verifying that Balance Sheet, Income Statement, and Cash Flow data align) before proceeding with scoring. This serves as a functional validation of data integrity.
Audit Metadata