suanli-nas-storage
Pass
Audited by Gen Agent Trust Hub on Sep 27, 2026
Risk Level: SAFECOMMAND_EXECUTIONDATA_EXFILTRATIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill performs API operations by executing local scripts included in the package.
- Evidence: It invokes
scripts/call.shandscripts/call_encrypt.pyvia the shell to communicate with the platform endpoints. - [DATA_EXFILTRATION]: The skill transmits user-provided API tokens, RSA private keys, and S3 credentials to the vendor's API endpoints to perform management tasks.
- Evidence: Credentials stored in environment variables or provided by the user are passed as arguments to the execution scripts, which send them to
openapi.suanli.cnandopenapi.suanleme.cn. - Mitigation: The instructions explicitly forbid the agent from logging these secrets or echoing sensitive S3 keys back to the user.
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process data from external API responses, which represents a theoretical attack surface.
- Ingestion points: Data returned from GET and POST requests to
openapi.suanli.cn(e.g., volume lists, task details). - Boundary markers: The skill does not define specific delimiters for API data, but relies on structured JSON responses.
- Capability inventory: The skill can create, modify, and delete storage volumes, and manage SFTP connections.
- Sanitization: The agent is instructed to use provided scripts to handle data transmission and encryption, reducing direct manipulation of sensitive payloads by the LLM.
Audit Metadata