The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes location names and metadata from crowd-sourced external services. This creates a surface where malicious instructions could be embedded in public map data.
Ingestion points: The scripts/maps_client.py script fetches data from nominatim.openstreetmap.org, overpass-api.de, and overpass.kumi.systems to provide location results.
Boundary markers: There are no explicit prompt boundary markers or instructions to the agent to ignore text embedded in the location data.
Capability inventory: The skill uses urllib.request to perform network operations to mapping service providers.
Sanitization: The skill does not validate or sanitize the string content returned from mapping APIs before presenting it to the agent context.

maps