The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill identifies an attack surface where data from Home Assistant entities (untrusted input) is ingested into automation logic. \n
Ingestion points: Home Assistant states are accessed via global.get("homeassistant").homeAssistant.states and various trigger nodes (e.g., trigger-state, events-state) as documented in SKILL.md and references/function-nodes.md. \n
Boundary markers: No explicit boundary markers are enforced within the provided JSON templates. \n
Capability inventory: The skill utilizes powerful capabilities including api-call-service (Home Assistant service calls), http request (outbound network requests), and ha-api (direct API access) as seen in SKILL.md and the references/ directory. \n
Sanitization: The references/security.md file provides clear instructions on sanitizing entity IDs and validating incoming payloads to prevent injection attacks. \n- [DYNAMIC_EXECUTION]: The skill facilitates the creation of function nodes that execute JavaScript at runtime, which is a core feature of the Node-RED platform. \n
Evidence: Multiple templates (e.g., templates/advanced-motion-light.json) and reference documents (e.g., references/function-nodes.md) contain JavaScript logic for processing automation events. \n- [COMMAND_EXECUTION]: The INSTALLATION.md file contains documentation for manual installation commands involving sudo and package managers. \n
Evidence: Commands such as sudo apt-get install -y nodejs and sudo npm install -g --unsafe-perm node-red are provided as instructional content for the user's manual setup, not for automated execution by the agent.

Node-RED