The Agent Skills Directory

[DATA_EXFILTRATION] (LOW): The script initiates network connections to rest.kegg.jp. While this is the legitimate endpoint for the KEGG database, any external network activity carries a baseline risk of data exposure if used to transmit sensitive information.
[INDIRECT_PROMPT_INJECTION] (LOW): The skill handles untrusted data from an external API.
Ingestion points: urllib.request.urlopen in all functions within scripts/kegg_api.py reads data from the KEGG REST API.
Boundary markers: None. The raw response is returned directly to the caller.
Capability inventory: No local write or execution capabilities are present in this script; it only performs GET requests.
Sanitization: None. Data is decoded as UTF-8 and returned.
[URL_MANIPULATION] (LOW): Several functions (e.g., kegg_info, kegg_list, kegg_conv) interpolate user-provided strings directly into the URL path without sanitization or URL encoding. This could allow an attacker to perform path traversal (e.g., using ../) to access unintended endpoints on the rest.kegg.jp server.

kegg-database