The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill fetches external book content and incorporates it into prompts for the LLM to generate social media threads. If the source content (fetched from jianshuo.dev) were to contain malicious instructions, the agent might inadvertently follow them when drafting tweets.
Ingestion points: daily.sh reads book content from today-material.txt, which is populated by scripts/fetch-book.py fetching from jianshuo.dev.
Boundary markers: The prompt uses plain text labels but lacks robust delimiters or explicit "ignore embedded instructions" warnings for the material content.
Capability inventory: The agent can execute the xurl tool to post content to the user's X/Twitter account.
Sanitization: fetch-book.py strips HTML tags but does not sanitize the text content for potential prompt injection patterns.
[COMMAND_EXECUTION]: The daily.sh script executes shell commands including curl for downloading cover images and xurl for interacting with the Twitter/X API to upload media and post tweets.
[PERSISTENCE_MECHANISMS]: The skill includes a launchd configuration (com.jianshuo.wjs-publishing-books-to-x.plist.template) to establish a daily scheduled task at 10:10 AM to run the publishing workflow.
[DYNAMIC_EXECUTION]: Shell scripts daily.sh and pick-next-book.sh use python3 - to execute inline Python code for parsing API responses, calculating character widths, and selecting the next book to process.
[EXTERNAL_DOWNLOADS]: The skill downloads book metadata, chapter content, and images from jianshuo.dev and voicedrop.cn. These resources are associated with the skill's stated purpose and the author's infrastructure.

wjs-publishing-books-to-x