The Agent Skills Directory

[PROMPT_INJECTION]: The author metadata provided in the skill manifest ('claude-code-skills') is inconsistent with the actual author context ('alirezarezvani'). This discrepancy is deceptive and could mislead users regarding the skill's origin.
[PROMPT_INJECTION]: The skill is designed to ingest and validate external markdown and JSON files using scripts such as kb_ingester.py and runbook_validator.py. This functionality exposes the agent to an indirect prompt injection surface where malicious content within processed documents could attempt to influence the agent's reasoning or instructions during a documentation review.
Ingestion points: The scripts kb_ingester.py, runbook_validator.py, and sop_generator.py read content from local files specified by the user.
Boundary markers: No delimiters or specialized instructions are used to separate the processed file content from the agent's core instructions.
Capability inventory: The Python scripts are limited to read-only file system operations and stdout reporting. No network or destructive capabilities are implemented.
Sanitization: While the scripts parse the structure of the documents, the text content is not filtered or sanitized for potential embedded instructions.
[SAFE]: The provided scripts (kb_ingester.py, runbook_validator.py, sop_generator.py) utilize only the Python standard library and do not require external dependencies, remote downloads, or network access.

knowledge-ops