The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes external source code to generate structural summaries (skeletons) for agent context. This creates an ingestion surface for potential indirect prompt injections embedded in the source code being analyzed.
Ingestion points: scripts/skeletonize_py.py (reads local file content) and scripts/skeletonize_ts.ts (processes source code strings).
Boundary markers: None implemented in the provided transformation logic.
Capability inventory: The scripts perform Abstract Syntax Tree (AST) manipulation and string output; they do not include subprocess execution, file writes, or network operations.
Sanitization: No filtering or sanitization of string literals or comments that might contain instructions is performed.

ast-skeletonizer