The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes a bash script wrapper (scripts/gmaps) that executes Node.js code to generate browser automation commands and pipes them to the browser-harness-js utility. It also executes a setup script to configure the environment.
[PROMPT_INJECTION]: The skill ingests unstructured data from Google Maps search results and route panels, representing an indirect prompt injection surface.
Ingestion points: Data enters via DOM scraping in the mapsSearch, directions, and resolvePlace functions within scripts/gmaps.
Boundary markers: The skill returns data to the agent in structured JSON or clearly defined text blocks.
Capability inventory: The skill possesses the ability to execute local shell commands and control browser targets via CDP.
Sanitization: Dynamic JavaScript payloads are constructed using JSON.stringify to escape user inputs and scraped data, preventing context escape within the browser execution environment.
[SAFE]: The skill's setup process involves adding $HOME/.local/bin to the system PATH by modifying shell profile files (~/.bashrc, ~/.zshrc, or ~/.profile), which is a standard and transparent method for installing local CLI tools.

gmaps