The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to use python3 subshells to inspect specific environment variables (prefixed with MNEMOSYNE) and to parse local YAML configuration files. These operations are limited to the agent's own configuration environment for troubleshooting purposes.
[EXTERNAL_DOWNLOADS]: The skill mentions external packages mnemosyne-memory and mnemosyne-hermes which are required for the memory provider's functionality. It references the official repository mnemosyne-oss/mnemosyne and provides instructions for repairing installations via the package's own installer script.
[PROMPT_INJECTION]: The skill manages a memory engine that ingests untrusted conversational data.
Ingestion points: Data enters the system via the working_memory database and episodic summaries created during the 'sleep' consolidation process.
Boundary markers: The instructions do not specify explicit delimiters for stored memory content but recommend using a host LLM for compression.
Capability inventory: The skill uses CLI tools (mnemosyne, hermes) and Python snippets for diagnostics and memory management.
Sanitization: The skill provides a detailed workflow for identifying and invalidating 'junk' or transport-layer artifacts (e.g., gateway errors) that could otherwise influence future agent responses.

hermes-mnemosyne