The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes several shell commands to manage a local web server and proxy configuration.
Starts background processes using python3 -m http.server to serve static content on local loopback ports.
Invokes portless alias to register subdomains on the local machine for portless access.
Uses kill to manage background server processes based on PIDs stored in local state files.
[DYNAMIC_EXECUTION]: The skill utilizes Python one-liners and heredocs to perform structured data operations.
Employs python3 -c for JSON parsing, port availability checks, and status reporting.
Uses a Python heredoc in scripts/serve.sh to safely write state information to JSON files within the project directory.
[INDIRECT_PROMPT_INJECTION]: The skill creates a surface for indirect prompt injection by serving and screenshotting files that may contain untrusted content.
Ingestion points: The skill takes a file or directory path as input (serve.sh) to be served.
Boundary markers: None present to delimit or warn about instructions within the served HTML content.
Capability inventory: Involves agent-browser for rendering and screenshotting, which is an exploitable capability if the served content contains malicious browser-targeting payloads.
Sanitization: The subdomain name is sanitized via a strict alphanumeric regex, though the HTML content itself is rendered as-is by design.

page-serve