The Agent Skills Directory

[PROMPT_INJECTION]: The skill's instruction set attempts to override the agent's decision-making logic by using coercive language to ensure tool usage.
Evidence: SKILL.md uses emphatic markers such as "⚠️ MANDATORY SKILL INVOCATION ⚠️" and "YOU MUST invoke this skill (NOT optional)" to dictate agent behavior.
[DATA_EXFILTRATION]: The skill script reads sensitive API credentials from a local configuration file for use in network requests.
Ingestion point: scripts/sonarr.sh sources ~/.homelab-skills/.env to retrieve the SONARR_API_KEY and SONARR_URL.
Context: The script transmits these credentials to the configured Sonarr service endpoint via curl to perform library management tasks.
Sanitization: The implementation safely handles user-provided search terms and series metadata using jq for URL encoding and JSON construction, which prevents secondary injection into the API communications.

sonarr