The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes the ngrok CLI to manage tunnels and uses curl to query the local ngrok agent API for public URLs.
[PERSISTENCE]: The skill generates a .ngrok/expose.sh script and adds a script entry to package.json, enabling users to easily re-activate the tunnel.
[INDIRECT_PROMPT_INJECTION]: The skill reads project configuration files to identify service ports. 1. Ingestion points: Reads package.json, .env, and docker-compose.yml to extract port values. 2. Boundary markers: None identified. 3. Capability inventory: ngrok CLI command execution, file writing, and package.json modification. 4. Sanitization: None; uses basic pattern matching.
[EXTERNAL_DOWNLOADS]: References official ngrok infrastructure for authentication and management, which are legitimate vendor resources.

expose-localhost