The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes local Python scripts (scripts/fetch_norikae_routes.py and scripts/build_norikae_url.py) using python3 to handle route planning logic and data retrieval.
[EXTERNAL_DOWNLOADS]: The skill connects to transit.yahoo.co.jp to fetch live train route information. This is a well-known transportation service in Japan.
[PROMPT_INJECTION]: The skill has an attack surface for indirect prompt injection as it processes external web content and presents it to the agent.
Ingestion points: External HTML/text content is ingested via scripts/fetch_norikae_routes.py (line 197).
Boundary markers: No explicit boundary markers or instructions to ignore embedded commands are present in the skill's workflow when processing fetched data.
Capability inventory: The skill possesses network access and the ability to execute its own local scripts.
Sanitization: Content is passed through strip_noise (to remove scripts and styles) and html_to_text functions before being processed by the agent.
[DATA_EXFILTRATION]: The fetch script includes a --url argument that allows fetching from any provided URL. This presents a Server-Side Request Forgery (SSRF) risk where the agent could be tricked into accessing internal services or metadata endpoints if a malicious URL is supplied.

norikae-guide