juanmackie

Mnemosyne Hermes — AI skill for Claude Code

AI community

Local-first persistent semantic memory and agent orchestration for Hermes, Claude Code, and MCP-compatible personal AI agents.

How to install Mnemosyne Hermes

This entry records only its repository, not the path inside it, so there is no exact command to give. Open juanmackie/mnemosyne-hermes and copy the folder into ~/.claude/skills/, or the file into ~/.claude/agents/.

What Mnemosyne Hermes does

Local-first persistent semantic memory and agent orchestration for Hermes, Claude Code, and MCP-compatible personal AI agents.

Alternatives in AI

Memmy Agent — 🍙 A personal AI agent & local memory hub for all AI agents, gives every AI one shared, fully controlled memor 1k ★
Iai Personal Memory Engine — A cyber brain for your AI 866 ★
Memtrace Public — Structural memory for AI coding agents 466 ★

README

Mnemosyne

**Fork notice**: This is a community-maintained fork of [rand/mnemosyne](https://github.com/rand/mnemosyne), optimized for the [Hermes](docs/HERMES_INTEGRATION.md) agent runtime — keyless personal-agent operation, graceful degradation without OS keyrings or cloud LLMs, Hermes MCP stdio contract compliance, and OpenViking-inspired hierarchical memory. All changes are released under the same MIT license.

**Current status (v2.4.0):** dynamic profile slice + typed `extends` edges are delivered and benchmarked; graph-aware `is_latest` is deferred to P5.

**Local-first persistent memory for Hermes and every MCP-compatible personal agent**

Mnemosyne provides private semantic memory with LibSQL vector search, full-text search, graph links, and hierarchical retrieval. It works without a cloud API key and keeps the standard MCP surface available for Claude Code, Cursor, Codex, Windsurf, and custom agents.

Hermes-first quickstart

# No Rust or Python required: downloads and verifies a native release binary.
curl -fsSL https://raw.githubusercontent.com/juanmackie/mnemosyne-hermes/main/install.sh | bash
export PATH="$HOME/.local/bin:$PATH"
hermes config set memory.provider mnemosyne

# Optional: preserve an existing Python mnemosyne-memory store.
mnemosyne import --from ~/.hermes/mnemosyne/data/mnemosyne.db \
  --namespace agent:hermes --dry-run --format json
mnemosyne import --from ~/.hermes/mnemosyne/data/mnemosyne.db \
  --namespace agent:hermes --format json

# Verify the local path without a cloud key.
unset ANTHROPIC_API_KEY OPENAI_API_KEY
mnemosyne remember --content "The user prefers local-only storage" \
  --namespace agent:hermes --no-enrich --format json

Then register `mnemosyne` in Hermes' `~/.hermes/config.yaml` under `mcp.servers` with `command: mnemosyne` and `args: ["mcp"]`. The complete install → configure → import → verify path is [docs/HERMES_INTEGRATION.md](docs/HERMES_INTEGRATION.md).