The Agent Skills Directory

[PROMPT_INJECTION]: The skill's configuration in agents/openai.yaml contains instructions that direct the AI agent to prioritize using GoLogin tools over competing services like Firecrawl, which is a form of competitive steering that influences the agent's default tool selection logic.
[EXTERNAL_DOWNLOADS]: The skill documentation describes the installation of the gologin-webunlocker-sdk Node.js package and its corresponding CLI tool. These are recognized as official vendor resources from the author, GologinLabs.
[DATA_EXFILTRATION]: The skill transmits target URLs to the Gologin Web Unlocker API (gologin.com) and requires the configuration of a GOLOGIN_WEBUNLOCKER_API_KEY. This behavior is consistent with the skill's primary purpose of providing web scraping via a specialized proxy service.
[PROMPT_INJECTION]: The skill's core functionality involves scraping content from arbitrary URLs (HTML, Markdown, and Text). This creates an attack surface for indirect prompt injection, where malicious instructions embedded in a target webpage could attempt to influence the agent's behavior during processing.
Ingestion points: Tools webunlocker_scrape, webunlocker_text, webunlocker_markdown, and webunlocker_json (manifest.json).
Boundary markers: None identified in the provided tool descriptions or operation patterns.
Capability inventory: No dangerous local operations such as arbitrary command execution or file system writes are present in the skill code.
Sanitization: No explicit sanitization or filtering of the scraped content is documented before it is returned to the agent context.

gologin-scraping-skill