The Agent Skills Directory

[COMMAND_EXECUTION]: The module utilizes subprocess.run() to execute git rev-parse --show-toplevel. This is a benign operation used to resolve the project's base directory for path construction. The command uses fixed arguments, preventing injection vulnerabilities.
[PROMPT_INJECTION]: The skill ingests untrusted data through markdown metadata parsing and JSONL storage helpers, creating a surface for indirect prompt injection if the files contain malicious instructions processed by a downstream agent.
Ingestion points: Markdown parsing in parse_markdown_metadata and JSONL reading in read_json and read_jsonl across scripts/file_ops.py and scripts/jsonl_store.py.
Boundary markers: None identified.
Capability inventory: The skill performs file system writes via atomic_write_file, creates directories with ensure_directory, and copies trees with copy_tree.
Sanitization: None; data processing relies on standard string operations and json.loads without additional content filtering.

tools-file-ops