The Agent Skills Directory

[SAFE]: The skill implements Server-Side Request Forgery (SSRF) protections in scripts/fetch_page.py and scripts/analyze_visual.py by resolving hostnames and blocking access to private, loopback, and reserved IP addresses.
[SAFE]: scripts/capture_screenshot.py includes directory traversal protection by validating that output paths are restricted to the current working directory or the user's home directory.
[DATA_EXFILTRATION]: The skill performs network operations to arbitrary user-provided URLs in scripts/fetch_page.py and scripts/analyze_visual.py. While SSRF mitigations are present, this capability allows for interaction with external, non-whitelisted domains.
[PROMPT_INJECTION]: The skill processes untrusted content from external websites, creating an attack surface for indirect prompt injection if malicious instructions are embedded in the analyzed pages.
Ingestion points: scripts/fetch_page.py (fetches raw HTML) and scripts/analyze_visual.py (renders content via Playwright).
Boundary markers: The skill converts raw HTML into structured data via scripts/parse_html.py before passing it to the agent, which serves as a data boundary.
Capability inventory: Use of the Bash tool and file system write operations in scripts/capture_screenshot.py.
Sanitization: Employs BeautifulSoup for structured HTML parsing rather than raw text interpolation.

seo