The Agent Skills Directory

[DATA_EXPOSURE]: The skill includes examples for configuring SSH tunnels that involve referencing local file paths for private keys (e.g., PrivateKeyPath), which is an expected configuration for the described tunneling functionality.
[COMMAND_EXECUTION]: The documentation describes libraries that manage and supervise external network tunneling agents such as cloudflared, ngrok, and tailscale to expose local services to the internet.
[INDIRECT_PROMPT_INJECTION]: The skill facilitates the ingestion of untrusted data through multiple protocol endpoints but includes descriptions of various security controls.
Ingestion points: SKILL.md (WebDAV, CalDAV, Tus, FileSync, and Webhook endpoints)
Boundary markers: Present (Webhook signature verification, Content-Digest integrity checks, and hardened XML parsing are explicitly mentioned)
Capability inventory: SKILL.md (File system writes to storage paths, ACME certificate management, and network tunneling operations)
Sanitization: Present (Containment checks for symlinks and duplicate delivery suppression for webhooks are detailed)

shiny-aspnet