The Agent Skills Directory

[SAFE]: The skill implements standard file management functionality for Enterprise WeChat using the official wecom-cli tool, which is a resource associated with the skill's authoring vendor.
[SAFE]: Privacy-preserving constraints are in place, explicitly instructing the agent not to expose internal local file paths or media identifiers to the user, following security best practices for data handling.
[SAFE]: Security guidelines are included to prevent the agent from attempting to bypass official encryption or anti-leaking policies (e.g., from work.weixin.qq.com), advising the agent to guide the user to the official client for protected resources instead.
[COMMAND_EXECUTION]: The skill invokes the wecom-cli binary to perform its core tasks. This execution is limited to specific JSON-structured arguments for uploading and downloading, which is the intended and documented purpose of the skill.
[INDIRECT_PROMPT_INJECTION]: The skill possesses a potential attack surface by ingesting external data.
Ingestion points: Local files read during the upload process and media identifiers provided for downloads.
Boundary markers: The instructions explicitly state the skill 'does not parse/identify file content', which significantly limits the potential for instructions embedded inside files to be processed by the agent.
Capability inventory: The skill executes shell commands via wecom-cli, performs local file writes during downloads, and reads local files during uploads.
Sanitization: The agent relies on the underlying CLI tool for protocol handling and follows the strict instruction to treat files as opaque blobs without parsing their content.

wecomcli-media