The Agent Skills Directory

[DYNAMIC_EXECUTION]: The scripts/chunk.py script performs dynamic module loading by modifying sys.path to import common client logic from a sibling skill directory. While this facilitates code reuse across the vendor's ecosystem, it relies on runtime path manipulation.
[COMMAND_EXECUTION]: The test suite in scripts/test_chunk.py utilizes subprocess.run to programmatically invoke the chunk.py script. This is used to validate CLI behavior and error handling within the testing environment.
[PROMPT_INJECTION]: The skill serves as a data ingestion gateway that processes document content for RAG. It does not implement sanitization or provide explicit boundary markers in its JSONL output, creating a potential surface for indirect prompt injection when the data is consumed by downstream LLMs.
Ingestion points: Document content is read from local files by scripts/chunk.py and processed via API.
Boundary markers: The output format lacks delimiters or instructions to treat the extracted text as untrusted data.
Capability inventory: The skill possesses capabilities for network communication, local file reading, and filesystem writing.
Sanitization: Document text is extracted and emitted without transformation or escaping.
[SAFE]: The skill adopts strong security measures for handling sensitive data, specifically using os.O_NOFOLLOW to prevent symlink attacks and enforcing 0600 file permissions on output artifacts containing extracted document text.

grounded-rag-ingestion