The Agent Skills Directory

[DATA_EXFILTRATION]: The skill instructions direct the agent to read sensitive API credentials from a local configuration file at ~/.gooseworks/credentials.json. While this is a standard authentication mechanism for the vendor's ecosystem, reading local credential files is a form of sensitive data exposure.
[COMMAND_EXECUTION]: The skill utilizes python3 command-line snippets to parse local JSON configuration files and recommends the execution of vendor-specific CLI commands for logging in.
[EXTERNAL_DOWNLOADS]: The documentation suggests the use of npx gooseworks login, which facilitates the download and execution of the vendor's command-line interface tool from a public registry.
[INDIRECT_PROMPT_INJECTION]: The skill's primary function is to scrape and process external web content (via website_url, website_html, or website_markdown) using AI models. This creates a surface for indirect prompt injection, where a malicious website could provide instructions that influence the agent's behavior.
Ingestion points: The website_url, website_html, and website_markdown parameters used in tools like SmartScraper and SmartCrawler.
Boundary markers: No specific delimiters or safety instructions are defined to separate untrusted web content from the agent's instructions.
Capability inventory: The skill possesses network communication capabilities via curl to the vendor's API endpoint at api.gooseworks.ai.
Sanitization: There is no evidence of content sanitization or filtering applied to the data retrieved from external URLs before it is processed by the AI.

ai-web-scraping-scrapegraph