volcengine-knowledge-search
Pass
Audited by Gen Agent Trust Hub on Oct 19, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill retrieves and processes content from external Volcengine documentation pages. This introduces a potential surface for indirect prompt injection, where malicious instructions hidden in the documentation could attempt to influence the agent's behavior. \n
- Ingestion points: Documentation content fetched via
ve docs searchandve docs fetchfromvolcengine.comdomains. \n - Boundary markers: The skill does not explicitly define prompt boundary markers or delimiters for the ingested text. \n
- Capability inventory: The skill utilizes the
veCLI tool for shell command execution. \n - Sanitization: While the CLI performs text cleaning, there is no explicit instruction for the agent to sanitize or escape the content before it is processed by the LLM.\n- [COMMAND_EXECUTION]: The skill invokes the
vebinary through shell commands, passing user-derived search queries as arguments. Standard quoting is used in the instructions to mitigate basic injection, but the agent must ensure robust argument handling during execution.\n- [EXTERNAL_DOWNLOADS]: The skill requires the installation of the@volcengine/clipackage. This is an official vendor resource provided by Volcengine.
Audit Metadata