The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests data from external JSON ledgers and interpolates these strings into generated Rust source files, creating a surface for indirect instructions to be placed in code comments.
Ingestion points: The scripts/emit_project.py script reads a JSON file (the ledger) which contains fields like id, detail, and evidence populated by previous analysis steps.
Boundary markers: No special delimiters or escaping mechanisms are used when placing ledger data into Rust documentation comments (///).
Capability inventory: The script possesses file-system write capabilities via Path.write_text and Path.write_bytes to generate the crate structure and copy assets.
Sanitization: The script includes a robust ident() function that sanitizes ledger IDs into safe Rust identifiers (alphanumeric and underscores), but it does not sanitize the descriptive strings interpolated into comments.

bevy-rea