mnemosyne-maintenance

Diagnose and repair Mnemosyne memory provider upgrades, embeddings, and import shadowing issues.

115|9|Updated Aug 5, 2026
One-click install
npx skills add https://github.com/AtlasOmnia/donna-starter --skill mnemosyne-maintenance-atlasomnia
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: mnemosyne-maintenance
Source: https://github.com/AtlasOmnia/donna-starter/tree/main/skills/hermes/mnemosyne-maintenance
Command: npx skills add https://github.com/AtlasOmnia/donna-starter --skill mnemosyne-maintenance-atlasomnia

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires mnemosyne-memory, sqlite-vec, fastembed.

What problem does it solve? Mnemosyne memory operations in Hermes can silently break after upgrades: consolidation hangs for hours, embeddings stay disabled despite installed dependencies, and version mismatches confuse diagnostics. This Skill provides a structured troubleshooting workflow to identify root causes and restore fast, vectorized memory operations. ## Core Features & Use Cases - Hung Consolidation Diagnosis: Detects when mnemosyne_sleep hangs due to missing embeddings on large databases and walks through enabling fastembed-based vectorization. - Import Shadowing Detection: Identifies when a local ~/mnemosyne directory or stale plugin symlink shadows the pip-installed package, causing wrong versions to load. - Embeddings Enablement: Step-by-step reindex workflow to vectorize existing memories with dry-run verification before writing. - Use Case: After upgrading Mnemosyne, mnemosyne_sleep hangs for 30+ minutes. Run the diagnostic sequence, discover a stale plugin symlink and missing MNEMOSYNE_VEC_TYPE env var, fix both, restart the gateway, and confirm consolidation completes in seconds. ## Quick Start Diagnose why my Mnemosyne memory consolidation is hanging and fix any version or embedding issues.

Frequently Asked Questions about mnemosyne-maintenance

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I fix mnemosyne_sleep hanging for a long time?

A hanging mnemosyne_sleep usually means embeddings are disabled, forcing brute-force FTS and LLM passes on large databases. Enable embeddings by setting MNEMOSYNE_VEC_TYPE=fastembed, installing sqlite-vec and fastembed, then running a reindex to vectorize existing memories.

Why does pip show a newer Mnemosyne version than the runtime reports?

A local ~/mnemosyne directory can shadow the pip-installed package because Python adds the home directory to sys.path on macOS. Verify with python -c "import mnemosyne; print(mnemosyne.__file__)" and rename the local copy if it resolves there.

How do I enable embeddings for Mnemosyne memory search?

Add MNEMOSYNE_VEC_TYPE=fastembed to ~/.hermes/.env, install sqlite-vec and fastembed, restart the gateway, then run mnemosyne reindex --dry-run followed by mnemosyne reindex --yes to vectorize existing memories.

Why does mnemosyne reindex fail with database not found errors?

The reindex command may default to ~/.mnemosyne/data/ instead of ~/.hermes/mnemosyne/data/. Set MNEMOSYNE_DATA_DIR=~/.hermes/mnemosyne before running, or use --no-backup if you are confident in database integrity.

Does a stale plugin symlink affect Mnemosyne imports?

Yes. A symlink at ~/.hermes/plugins/mnemosyne pointing to a local dev copy breaks imports on modern Mnemosyne 3.8+, which bundles Hermes integration directly. Remove the symlink and restart the gateway to resolve it.