The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes local Python scripts (kb_ingester.py, runbook_validator.py, sop_generator.py) to automate documentation audits. These scripts rely exclusively on the Python standard library and do not perform network exfiltration or high-risk system operations.
[INDIRECT_PROMPT_INJECTION]: The tools scan local markdown files and JSON metadata, which constitutes a surface for indirect prompt injection.
Ingestion points: kb_ingester.py recursively reads markdown files within a target directory to extract structural metadata. runbook_validator.py and sop_generator.py process individual input files.
Boundary markers: No explicit delimiters or instructions are used to isolate document content from the agent's control flow.
Capability inventory: The skill allows the agent to read local documentation and receive metadata reports (e.g., titles, glossary terms) in its context.
Sanitization: Content extraction is performed through regex targeting specific fields like YAML frontmatter and markdown headers, which reduces but does not eliminate the risk of the agent interpreting embedded instructions.

knowledge-ops