The Agent Skills Directory

[DYNAMIC_EXECUTION]: The script scripts/mcp_server.py performs dynamic loading of internal modules for its operations.
Evidence: It uses importlib.util.spec_from_file_location to load the vault_index.py module from a relative path within the skill's internal directory structure.
[INDIRECT_PROMPT_INJECTION]: The skill provides tools for reading and processing local Markdown files, which represents a potential attack surface for indirect prompt injection.
Ingestion points: Files stored within the user-defined VAULT_DIR are read via tools like read_kb_note in scripts/mcp_server.py.
Boundary markers: The current implementation does not use explicit boundary markers or delimiters when returning file content to the agent.
Capability inventory: The skill includes file reading, SQLite FTS5 indexing, and directory scanning capabilities.
Sanitization: Content is retrieved directly from the file system and passed to the agent without escaping or validation filters.
[EXTERNAL_DOWNLOADS]: The skill configuration and documentation specify the installation of external Python packages from standard registries.
Evidence: requirements-mcp.txt lists the mcp library (v1.30.0), which is a well-known package for implementing the Model Context Protocol.
[COMMAND_EXECUTION]: The skill instructions guide the user to execute Python scripts via the command line for vault maintenance.
Evidence: SKILL.md describes how to initialize and search the knowledge base using tools/chubby.py, scripts/vault_health_check.py, and scripts/mcp_server.py.

knowledge-base-management