The Agent Skills Directory

[SAFE]: The skill is designed for local article processing. It identifies a clear workflow for cleaning Markdown files, generating titles/summaries, and interacting with the article-image skill for visuals.
[DATA_EXPOSURE]: Analysis of file operations confirms the skill only accesses the .article-skills/ directory and the project root for its specific outputs. No access to sensitive system paths (~/.ssh, .env, etc.) was detected. The documentation in auto-publish.md explicitly advises against storing credentials in Markdown files, demonstrating a security-conscious design.
[COMMAND_EXECUTION]: The skill uses structured tool calls (AskUserQuestion, Skill) for its operations. There are no instances of arbitrary shell command execution or subprocess spawning.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted article content as its primary input. While it lacks explicit boundary markers to separate data from instructions during the cleaning phase, its restricted capabilities (writing to specific local paths and generating text) mitigate the risk of exploitation.
[DYNAMIC_CONTEXT_INJECTION]: No use of the !command syntax was found in the SKILL.md file.

article-publish