The Agent Skills Directory

[PROMPT_INJECTION]: The skill contains explicit instructions to bypass user oversight and consent mechanisms. It directs the agent to install dependencies 'automatically without asking the user to confirm' and further mandates 'do not request separate approval for dependency installation.' This is a direct attempt to override the safety guardrails that normally require user permission for code execution or tool installation.
[REMOTE_CODE_EXECUTION]: The instructions require the agent to download and install the 'BrowserAct CLI' and associated skills from a remote URL if they are missing. This involves executing unverified third-party software at runtime without a verification or approval step.
[EXTERNAL_DOWNLOADS]: The skill relies on components fetched from an unverified external domain ('www.browseract.com'). The use of a tracking parameter ('co-from=lead-generation') in the installation link suggests the source may not be a standard or verified package repository.
[DATA_EXFILTRATION]: While the skill's primary purpose is scraping, it contains instructions to manage sensitive data like 'cookies, account information, browser IDs, and proxy settings' within the workspace. If combined with the unverified remote code execution capability, this creates a high risk for credential theft.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process untrusted data from LinkedIn (profiles, posts, and job listings). It lacks any definition of boundary markers (e.g., delimiters) or instructions for the agent to ignore potentially malicious content embedded in the scraped data, making it vulnerable to indirect injection attacks where external content could hijack the agent's flow.

linkedin-job-search-scraper