The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process data from external codebases using search and read tools. There is an inherent risk of indirect prompt injection if a codebase contains malicious instructions embedded in comments or documentation. While the subagents are restricted to readonly mode, these instructions could potentially influence the resulting architectural explanation or critique.
Evidence: The prompt templates in references/explorer-prompt.md and references/explainer-prompt.md instruct subagents to read and trace code implementations without explicit sanitization or boundary markers for untrusted content.
[DYNAMIC_EXECUTION]: The skill implements a complex orchestration flow that dynamically spawns multiple subagents (Explorers, Explainers, and Critics) using specialized prompts.
Evidence: SKILL.md contains logic to spawn generalPurpose subagents with specific model configurations and prompt templates based on the complexity of the user's request.

how