bailian-web-search
Pass
Audited by Gen Agent Trust Hub on Sep 11, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [COMMAND_EXECUTION]: The skill utilizes the 'bl' command-line interface to manage routing logic and execute searches. Besides, the skill uses these commands to perform identity detection by checking the base_url host against expected Alibaba Cloud domains.
- Evidence: The skill invokes
bl config show,bl text chat, andbl search webto identify identity and perform search operations withinSKILL.md. - [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from web search results and passes it into the agent's context, creating an indirect prompt injection surface.
- Ingestion points: User-provided queries and messages are passed to the
--messageand--queryflags inSKILL.md. - Boundary markers: No explicit delimiters or boundary markers for external content are defined within this file.
- Capability inventory: The skill has command execution and network access capabilities through the 'bl' tool.
- Sanitization: There is no evidence of sanitization for the search results returned by the CLI.
- [EXTERNAL_DOWNLOADS]: The skill documentation refers to an official installation method that involves downloading resources from the vendor's infrastructure.
- Evidence: The skill instructions specify using
bl skill initfor installation and setup as described inSKILL.md.
Audit Metadata