The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill serves as a router that directs the agent to ingest external instructions by loading other skill documents.
Ingestion points: The agent is instructed to use rig context get <ref> to pull the body content of other skills into its active context.
Capability inventory: The skill leverages the Bash(rig:*) tool to interact with the platform's context management system.
Boundary markers: The document outlines a "progressive disclosure" model (hot tier vs. body) to manage how much information is loaded at once.
Sanitization: The skill relies on the underlying rig CLI to serve content from the local installation, ensuring that loaded skills match the running version.
[COMMAND_EXECUTION]: The skill provides instructions and examples for using the platform-specific rig CLI tool to list and retrieve context.
Evidence: Examples include rig context list and rig context get skills/core/rig-lifecycle.
Context: These commands are restricted by the allowed-tools frontmatter and are standard operations for the OpenRig environment.

openrig-skills