The Agent Skills Directory

[DATA_EXPOSURE_AND_EXFILTRATION]: The skill uses an environment variable RUNTIME_API_KEY for authentication and performs network requests via curl to the vendor's infrastructure (runtm.com). This is standard procedure for an API-integrated skill and does not represent an unsafe exfiltration pattern.
[INDIRECT_PROMPT_INJECTION]: The skill defines a data ingestion surface where external API output is piped into jq and potentially processed by the agent (found in SKILL.md). There are no specific boundary markers or sanitization steps mentioned to handle potentially malicious instructions in the API response, representing a common indirect prompt injection surface.

runtime