The Agent Skills Directory

[COMMAND_EXECUTION]: The skill directs the agent to use the iris CLI for provisioning and managing intent router services. These commands are typical administrative actions within the Redis Iris ecosystem.
[INDIRECT_PROMPT_INJECTION]: The skill implements a semantic routing surface that processes external input to determine agent behavior.
Ingestion points: User messages are passed to the router.match() or router.amatch() methods to determine the intended route.
Boundary markers: The skill does not define specific delimiters or instructions to the LLM to ignore embedded commands within the message string.
Capability inventory: The classification outcome determines which tool, sub-agent, or prompt is invoked next.
Sanitization: The input is processed semantically by the Redis Iris service; no local sanitization or filtering of the input string is performed by the skill code.
[SAFE]: The skill demonstrates proactive security measures by instructing the agent to avoid exposing API keys in terminal transcripts. It mandates the use of secure CLI flags like --write-env for secret storage and explicitly prohibits using commands like cat or grep on sensitive configuration files.

add-intent-router