The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute shell commands using node scripts/call.mjs and curl. These commands are used to query geocoding services, search catalogs, and retrieve service schemas.- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user data, specifically street addresses and geographic coordinates, which are interpolated directly into shell command arguments for scripts/call.mjs. This creates an attack surface where maliciously crafted addresses could attempt to influence the shell command execution.
Ingestion points: User-provided strings for address, addresses, lat, lon, and radius parameters found in the geocoding recipes in SKILL.md.
Boundary markers: The documentation advises the agent to write addresses without internal commas to prevent mis-splitting and to use dry runs to verify the request body before payment.
Capability inventory: The skill utilizes node and curl for external interactions and payment processing.
Sanitization: The provided documentation does not detail input sanitization procedures, relying on the agent to follow formatting guidelines.- [EXTERNAL_DOWNLOADS]: The skill communicates with the author's domain https://fittings.sh to fetch catalog information and execute geocoding API calls. These interactions are documented as the core functionality of the skill.

fittings-geo