The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill automatically downloads and installs the Playwright framework and Chromium browser binaries using the command npx -y playwright install chromium if the dependencies are not detected at runtime.
[COMMAND_EXECUTION]: Uses the child_process.spawn method to execute shell commands for installing browser dependencies, with specific handling for Windows (cmd.exe /c) and Unix-like environments.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process arbitrary data from external WeChat articles, creating a vulnerability surface where malicious content could attempt to influence agent behavior.
Ingestion points: The scripts/fetch.js script fetches article content from user-provided URLs using Playwright.
Boundary markers: The instructions do not define any delimiters or provide a "ignore embedded instructions" warning to separate fetched article content from the agent's instructions.
Capability inventory: The skill possesses network access (Playwright, https/http), file system write capabilities, and the ability to trigger the execution of other skills in the environment.
Sanitization: While the script performs HTML tag removal to extract text, it lacks sanitization or filtering to detect or neutralize embedded natural language instructions within the article text.
[PROMPT_INJECTION]: The instructions contain a "Mandatory Execution Protocol" that explicitly directs the agent to override standard interaction patterns by automatically invoking the legal-text-format skill without seeking user confirmation ("必须立即调用 legal-text-format，不询问用户确认").

wechat-article-fetch