longbridge-research
Pass
Audited by Gen Agent Trust Hub on Oct 12, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill uses web search as a fallback to gather data from external sources (e.g., DefiLlama, CoinGecko, HKEXnews, and general market news) when the internal CLI tool cannot provide specific metrics like DeFi yields or on-chain data. This allows untrusted external content to enter the agent's context.
- Ingestion points: External data is fetched via web search in
references/defi-yield.md,references/onchain.md,references/hkipo-analysis.md, andreferences/finance-calendar.md. - Boundary markers: The provided instructions do not specify any boundary markers or special handling to ensure the agent ignores potentially malicious prompts contained within the retrieved external text.
- Capability inventory: The skill can execute various
longbridgeCLI commands and access sensitive user data such as portfolio values and positions (references/financial-planning.md). - Sanitization: There is no explicit logic described to sanitize or validate the external data before it is presented to the agent or the user.
Audit Metadata