The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes retrieved memory data via qdrant_memory_search (defined in SKILL.md and references/tools.md) which could contain malicious instructions. However, the skill provides clear defensive guidance for the agent to treat this content as data only and never as commands. It also instructs the agent to verify retrieved context against current evidence.
[COMMAND_EXECUTION]: The bundled scripts/install_runtime.py uses subprocess.run to install dependencies and verify the environment. This is a standard installation procedure and includes extensive checks to ensure the installation path is private, user-owned, and secure against path injection or unauthorized access (e.g., _require_private_parent and path character validation).
[EXTERNAL_DOWNLOADS]: The installer fetches the hermes-plugin-qdrant-memory package from PyPI or the author's official GitHub repository. These are vendor-owned resources (cnkang) and their installation is gated by explicit user authorization as described in references/runtime.md.

qdrant-memory