The Agent Skills Directory

[SAFE]: The skill uses standard system commands (open, xdg-open) and standard library modules (http.server) to perform tasks consistent with its stated purpose of website previewing. No malicious obfuscation, exfiltration, or persistence patterns were found.
[PROMPT_INJECTION]: The skill has an indirect prompt injection surface as it is designed to ingest and process content from a generated website directory.
Ingestion points: HTML files and other assets within output/website/ (referenced in SKILL.md).
Boundary markers: Absent; no specific instructions are provided to the agent to isolate the website content from its own instruction context.
Capability inventory: The agent has access to Bash (used in SKILL.md for open and xdg-open commands) and Chrome automation tools (mcp__claude-in-chrome in SKILL.md frontmatter).
Sanitization: Absent; the skill does not specify any sanitization for the data read from the local files.
Note: This interaction is expected for a website validation tool and does not indicate malicious intent.

website-preview