The Agent Skills Directory

[DYNAMIC_EXECUTION]: The skill generates and executes a Node.js HTTP server script at runtime. The instructions provide a JavaScript template for an HTTP service that is written to a file and executed in the background to facilitate data storage, which bypasses context window limitations and static analysis of the runtime environment.
[COMMAND_EXECUTION]: The skill uses shell commands with the run_in_background: true flag to start a persistent local Node.js service and perform file system operations via generated scripts.
[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from external Weibo search results and user posts, which could contain instructions targeting the agent or the Markdown consumer.
Ingestion points: External Weibo content scraped from s.weibo.com in steps 2 and 4.
Boundary markers: None identified; scraped content is placed directly into Markdown blocks without clear delimiters or instructions to ignore embedded commands.
Capability inventory: The skill possesses file-write capabilities (fs.writeFileSync), background process execution, and broad browser automation tools.
Sanitization: There is no explicit sanitization, filtering, or escaping mentioned for the external content before it is processed into final documents.
[REMOTE_CODE_EXECUTION]: The skill utilizes the browser_run_code_unsafe tool to execute arbitrary, skill-defined JavaScript within the browser context to extract data and interact with the local persistence server.
[EXTERNAL_DOWNLOADS]: The skill instructs the user to install and trust the @playwright/mcp package. This is a reference to a well-known service and is documented here for completeness.

weibo-hot-scraper