feishu-send-file
Fail
Audited by Gen Agent Trust Hub on Aug 19, 2026
Risk Level: HIGHCOMMAND_EXECUTIONREMOTE_CODE_EXECUTIONPROMPT_INJECTION
Full Analysis
- [DYNAMIC_EXECUTION]: The SKILL.md file contains a usage example recommending the use of Python's exec() function to run shell commands with interpolated variables. This pattern is unsafe as it allows for command injection if the variables are not properly sanitized. The documentation also provides examples using python3 -c to process JSON output from curl commands.
- [UNVERIFIABLE_DEPENDENCIES_AND_REMOTE_CODE_EXECUTION]: The scripts scripts/send_file.py and scripts/send_image.py use subprocess.run() to invoke the system curl binary for uploading files to remote servers.
- [UNVERIFIABLE_DEPENDENCIES_AND_REMOTE_CODE_EXECUTION]: The skill performs network operations to Feishu and Lark API domains (open.feishu.cn and open.larksuite.com) for authentication and messaging. These are identified as well-known enterprise communication services.
- [INDIRECT_PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface due to its ingestion of local files and user-provided metadata for transmission to external services. 1. Ingestion points: File contents and command-line arguments (file paths, names, IDs) processed by the Python scripts. 2. Boundary markers: None identified. 3. Capability inventory: Execution of curl via subprocess and network requests via urllib. 4. Sanitization: No sanitization or validation of input paths or metadata is performed before their use in command execution or API payloads.
Recommendations
- HIGH: Downloads and executes remote code from: https://open.feishu.cn/open-apis/im/v1/messages?receive_id_type=open_id, unknown (check file) - DO NOT USE without thorough review
Audit Metadata