The Agent Skills Directory

[COMMAND_EXECUTION]: The skill invokes the bl CLI to retrieve current configuration details (bl config show) and perform web searches (bl search web, bl text chat). These operations are core to the skill's purpose of distributing search tasks across different Alibaba Cloud backends.
[INDIRECT_PROMPT_INJECTION]: The skill processes user-provided search strings by interpolating them into shell command arguments, creating a surface for potential injection.
Ingestion points: User queries are inserted into the --message and --query flags in the SKILL.md templates.
Boundary markers: The instructions do not define specific delimiters or warning markers to prevent the agent from following instructions embedded within the search query.
Capability inventory: The skill has the capability to execute shell commands using the bl tool across multiple routing paths.
Sanitization: No explicit logic for sanitizing or escaping user-supplied input is present in the skill instructions.
[DYNAMIC_EXECUTION]: The skill dynamically constructs and executes shell commands based on runtime configuration detection, specifically deciding between model-native and MCP paths based on the active profile and hostname.

bailian-web-search