The Agent Skills Directory

[PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface as it is designed to monitor external data sources such as files, systems, metrics, and Home Assistant sensor states. * Ingestion points: Data enters the agent context from monitored files, system metrics, or Home Assistant entities (defined in SKILL.md). * Boundary markers: The procedure requires the agent to restate target boundaries (Step 1) and implements a mandatory stop for explicit human confirmation before taking live actions (Step 7). * Capability inventory: The skill uses curl to interact with the Home Assistant REST API (referenced in Prerequisites and Procedure). * Sanitization: While technical sanitization of input data is not explicitly defined, the manual confirmation gate serves as a significant mitigation against automated exploitation of the monitoring logic.- [SAFE]: Credentials and sensitive access tokens (e.g., HA_TOKEN) are managed through environment variables rather than being hardcoded, which aligns with secure configuration standards.- [SAFE]: The skill uses curl for legitimate integration with the Home Assistant API as part of its primary stated purpose, with no evidence of unauthorized data exfiltration or connections to untrusted domains.

glyph-of-warding