alibabacloud-find-skills
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADSDYNAMIC_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied search keywords and intent phrases, which are then interpolated into API requests and shell commands. This creates a surface where external data can influence agent behavior.
- Ingestion points:
SKILL.md(Search Skills workflow) - Boundary markers: Use of
--data-urlencodefor HTTP parameters provides some isolation for the network request. - Capability inventory: Shell command execution (Bash/PowerShell), network operations (
curl), and package installation (npx). - Sanitization: Instructions recommend URL encoding for query parameters but do not specify shell escaping for the provided
powershell -Commandorbashtemplates. - [COMMAND_EXECUTION]: The skill provides pre-formatted shell command templates for Bash and Windows PowerShell that include placeholders for user-supplied keywords. If the agent fails to properly escape the user's input before placing it in these templates, it could lead to command injection on the host system.
- [EXTERNAL_DOWNLOADS]: The skill performs network requests to
https://agentexplorer.aliyuncs.comto retrieve category lists and skill metadata. It also facilitates the installation of additional skills from remote repositories usingnpx. These operations target the official infrastructure of the verified vendor. - [DYNAMIC_EXECUTION]: The skill uses
powershell -Commandto execute dynamically constructed command strings on Windows environments to maintain cross-platform compatibility for its discovery and installation workflows.
Audit Metadata