The Agent Skills Directory

[SAFE]: The skill contains high-level instructions and best practices for token management and context optimization. It defines strategies for routing verbose data to subagents to preserve reasoning quality in the main thread.
[COMMAND_EXECUTION]: The skill provides a pre-defined awk command template for parsing text files and extracting specific numbered list sections. This command is a static utility intended for text filtering and does not introduce risk when used as instructed.
[INDIRECT_PROMPT_INJECTION]: The skill describes procedures for processing external data such as large payloads, command outputs, and playbook files.
Ingestion points: External playbook files, screenshots, and command outputs (e.g., grep, jq, awk results) in SKILL.md.
Boundary markers: None explicitly defined within these instructions.
Capability inventory: File reading, shell command execution, and subagent spawning identified across the guidance.
Sanitization: None specified; however, the skill focuses on reducing the volume of raw data ingested into the main context window, which indirectly reduces the exposure to large-scale injection attacks.

principle-guard-the-context-window