The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted codebase content which could contain malicious instructions designed to influence agent behavior.
Ingestion points: As defined in references/explorer-prompt.md, subagents use Glob, Grep, and Read to ingest content from the target codebase.
Boundary markers: Collected code snippets and findings are interpolated directly into templates (e.g., {EXPLORER_FINDINGS_ALL}, {EXPLANATION}, and {FILE_PATHS}) without specialized delimiters or 'ignore' instructions for embedded content.
Capability inventory: The skill relies on read-only file system tools (Read, Grep, Glob) and does not request network access or persistent system modifications.
Sanitization: No escaping, validation, or sanitization is applied to the ingested code before processing by the AI agents.

how