The Agent Skills Directory

[CREDENTIALS_UNSAFE]: The skill requires the user to provide an "XHS_COOKIE" in an ".env" file. This cookie contains session tokens that grant full access to the user's Xiaohongshu account. Evidence found in publish_xhs.py and publish_xhs_browser.py.
[DATA_EXFILTRATION]: The script scripts/publish_xhs.py features an "API mode" that sends the user's session cookie to a configurable server defined by the XHS_API_URL environment variable. While it defaults to localhost, this mechanism creates a vector for sending sensitive credentials to arbitrary network locations.
[EXTERNAL_DOWNLOADS]: The skill fetches browser binaries via the Playwright framework and loads external typography from Google Fonts (fonts.googleapis.com) during the image rendering process. These are well-known services used for the skill's primary purpose.
[COMMAND_EXECUTION]: The skill's setup instructions require the execution of playwright install chromium to download and install browser components on the host system.
[PROMPT_INJECTION]: The skill possesses a surface for indirect prompt injection as it processes untrusted user-provided Markdown content into HTML templates for rendering.
Ingestion points: Markdown content provided by the user (processed in scripts/render_xhs.py and scripts/render_xhs.js).
Boundary markers: None identified in the prompt templates.
Capability inventory: Network operations in publish_xhs.py and local file access through the Playwright browser engine.
Sanitization: Relies on standard Markdown-to-HTML conversion which may not strip all malicious payloads.

xhs-note-creator