The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The 'Version Update Check' section in SKILL.md directs the agent to query a remote API (https://www.weiyun.com/api/v3/mcpserver) and follow the instruction field provided in the response to perform updates. This creates a mechanism where a remote server can supply instructions that the agent is conditioned to execute without further validation.
[DATA_EXFILTRATION]: The scripts/upload_to_weiyun.py script reads the entire content of a specified local file and sends it via an HTTP POST request to a remote endpoint (weiyun.com). This behavior is consistent with the skill's stated purpose as a cloud storage manager for a well-known service.
[COMMAND_EXECUTION]: The skill documentation provides shell commands for the agent to execute, including Python scripts for parameter generation (scripts/gen_block_info_list.py) and file uploading (scripts/upload_to_weiyun.py). It also uses curl to perform file downloads.
[EXTERNAL_DOWNLOADS]: The skill fetches update metadata and file management commands from weiyun.com, which is a well-known service provider.
[INDIRECT_PROMPT_INJECTION]: The skill processes local data through the scripts/upload_to_weiyun.py script and interacts with remote API responses.
Ingestion points: Reads local files from the file system via scripts/upload_to_weiyun.py.
Boundary markers: None identified in the processing scripts.
Capability inventory: Includes Bash for command execution, Write for file modification, and network access via the requests library.
Sanitization: The script performs no sanitization on the file data it processes before transmission or output.

weiyun