The Agent Skills Directory

[COMMAND_EXECUTION]: The script scripts/format.py uses subprocess.Popen to open the generated HTML article in the user's default web browser for previewing purposes.
[COMMAND_EXECUTION]: The publishing script scripts/publish.py executes the formatting script format.py as a subprocess to ensure the content is correctly prepared before transmission to WeChat.
[DATA_EXFILTRATION]: The skill transmits article text and image data to the official WeChat API (api.weixin.qq.com) to create draft posts. This is the primary intended function of the skill.
[DATA_EXFILTRATION]: AI-powered components in scripts/comment_reply.py and scripts/generate.py send article snippets or image prompts to external AI gateways (such as OpenRouter or Gemini-compatible services) to generate content and replies.
[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted Markdown content from user-provided files to perform structural analysis and automated styling.
Ingestion points: User-provided article files read in scripts/format.py and scripts/publish.py.
Boundary markers: The SKILL.md includes explicit instructions for the AI to only add structure markers and not modify the underlying content.
Capability inventory: File system read/write access, network requests to WeChat and AI provider APIs, and local subprocess execution.
Sanitization: The formatting engine uses standard HTML escaping for text interpolation in its gallery and preview templates.

hkr-render