The Agent Skills Directory

[CREDENTIALS_UNSAFE]: The script scripts/clip_wechat_tikhub.py utilizes the winreg module to read from the Windows registry at HKEY_CURRENT_USER\Environment to retrieve the TIKHUB_API_KEY credential.
[EXTERNAL_DOWNLOADS]: The skill performs network requests to api.tikhub.io to download article metadata and content using the urllib library.
[DATA_EXFILTRATION]: The skill transmits the user's TIKHUB_API_KEY to the third-party domain api.tikhub.io via the Authorization header in all API requests.
[PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection through the processed article content.
Ingestion points: External article content is ingested from the TikHub API in scripts/clip_wechat_tikhub.py.
Boundary markers: Absent. The instructions in SKILL.md and the output generation logic do not wrap fetched content in delimiters or provide instructions to ignore natural language commands embedded in the articles.
Capability inventory: The skill combines the ability to read untrusted data from the network with the ability to write that data to the local filesystem.
Sanitization: The script performs structural cleaning of HTML and WeChat UI elements but does not filter for natural language instructions or malicious prompts within the article body text.

wechat-clippings-vp