The Agent Skills Directory

[CREDENTIALS_UNSAFE]: The script scripts/heiyan-booklist-scraper.js contains logic to programmatically extract session credentials. It uses document.cookie.match(/Admin-Token=([^;]+)/) to retrieve the 'Admin-Token' from the user's active browser session, which is then used to authenticate subsequent API requests to the platform.
[COMMAND_EXECUTION]: The utility script scripts/cdp-utils.js utilizes child_process.execSync to run the agent-browser CLI tool. This allows the skill to execute arbitrary shell commands on the host machine to manage the browser automation environment.
[DYNAMIC_EXECUTION]: The skill makes extensive use of eval capabilities through the CDP interface. Both scripts/dz-browse-scraper.js and scripts/heiyan-booklist-scraper.js execute dynamically constructed JavaScript strings within the context of the automated browser to scrape data and interact with page elements.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted content from external novel platforms, creating a potential surface for indirect prompt injection.
Ingestion points: Data is scraped from ishugui.com and manage.zhangwenpindu.cn via scripts/dz-browse-scraper.js and scripts/heiyan-booklist-scraper.js.
Boundary markers: None identified in the prompt instructions provided in SKILL.md to distinguish between scraped data and agent instructions.
Capability inventory: The skill has access to execSync in scripts/cdp-utils.js and file system write access (fs.writeFileSync) in both scraping scripts.
Sanitization: There is no evidence of sanitization or escaping of the scraped novel titles, descriptions, or tags before they are presented to the agent for analysis.

story-short-scan