The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses the playwright-cli tool to perform browser automation tasks. It provides JavaScript snippets to be executed within the browser context for tasks such as dismissing cookie banners, navigating search results, and extracting data using page.evaluate().
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it ingests and processes untrusted data from an external website (jobdanmark.dk).
Ingestion points: Data is extracted in SKILL.md (Step 3: Extract Job Listings) using document.querySelectorAll('a.job-list') and page.evaluate().
Boundary markers: The instructions do not define specific delimiters or warnings to the agent regarding potential instructions embedded within job listings.
Capability inventory: The skill has access to Bash(playwright-cli:*) which allows it to execute arbitrary shell commands and browser automation code.
Sanitization: There is no evidence of sanitization or filtering of the extracted job titles, company names, or location data before it is presented to the agent.
[SAFE_PRACTICES]: The documentation includes explicit security guidance for the agent to avoid shell injection vulnerabilities, specifically cautioning against the use of the & character in bash strings and recommending URLSearchParams or Unicode escapes as safe alternatives.

jobdanmark-search