The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill uses npx -y socialdatax-skills@latest to download and execute code directly from the NPM registry at runtime. While this is functional for its purpose, it introduces a dependency on an external package that is not pinned to a specific version.
[COMMAND_EXECUTION]: The skill executes shell commands using node to run local scripts (src/xiaohongshu/post-cli.js) and npx for remote tools. These commands incorporate user-provided URLs as arguments, which could lead to command injection if the agent environment does not properly sanitize the input.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it ingests and processes untrusted data from Xiaohongshu (titles, descriptions, and comments).
Ingestion points: Data enters via the xiaohongshu-content-tools and xhs-content-research outputs (specifically post titles and content summaries).
Boundary markers: None explicitly defined to separate the fetched content from the analytical instructions.
Capability inventory: The skill can execute shell commands and read/write to the local file system (logs).
Sanitization: There is no evidence of filtering or escaping external content before the agent analyzes it.

space-xhs-account-audit