The Agent Skills Directory

[SAFE]: No external dependencies are used; both scripts provided in the skill rely exclusively on the Python standard library.
[SAFE]: The analysis detected no network operations, subprocess executions, or dynamic code evaluation patterns within the provided Python scripts.
[PROMPT_INJECTION]: The skill implements a retrieval pipeline for untrusted markdown notes, which creates an attack surface for indirect prompt injection.
Ingestion points: The scripts scripts/scan_vault.py and scripts/build_context_pack.py read and process all markdown files within a user-defined vault directory.
Boundary markers: The packaging script (build_context_pack.py) uses structured markdown formatting, including headers, horizontal rules, and blockquoted metadata (source path, modification time), to delineate individual notes and metadata from their content.
Capability inventory: The skill's scripts are restricted to file system read and write operations; they lack capabilities for network exfiltration, arbitrary command execution, or persistence.
Sanitization: The tool interpolates raw note content into context packs without escaping or filtering, relying on the boundary markers and agent-side guardrails to manage potentially malicious instructions embedded in notes.

docs-notes-retrieval