hermes-memory-providers
Pass
Audited by Gen Agent Trust Hub on Aug 28, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill implements a memory layer that ingests conversation history and injects it back into the agent's system prompt, creating a potential surface for indirect prompt injection.\n
- Ingestion points: Conversation turns are automatically stored to episodic memory via the
post_turn_synclifecycle hook as described inSKILL.md.\n - Boundary markers: Memories are injected into the agent's context using a
# Mnemosyne Memoryblock header.\n - Capability inventory: The plugin registers 20 tools (e.g.,
remember,recall,sleep,triples,graph) and CLI commands for managing stored data.\n - Sanitization: No specific sanitization or filtering logic for stored memory content is mentioned in the provided installation instructions.\n- [EXTERNAL_DOWNLOADS]: The skill requires the installation of the
mnemosyne-hermesandmnemosyne-memoryPython packages from external registries. These represent the primary functionality of the memory provider developed by the author.\n- [COMMAND_EXECUTION]: The installation process involves executing shell commands to link the plugin (mnemosyne-hermes install) and configure the Hermes environment.
Audit Metadata