Memanto: memory tool for Hermes Agent

Memory agent that curates, reconciles and shares long-term memory across multiple AI agents

What Memanto does

Memanto describes itself as a companion agent rather than an API. It observes agent interaction streams and extracts durable knowledge such as decisions, preferences, facts and failures, merges duplicates into one canonical store, and supersedes old facts when new ones contradict them while preserving what was believed and when. Decay, expiry and deletion run as policies, and before an agent acts Memanto hands it a minimal relevant slice of the stored memory.

Setup is a pip install followed by an interactive command that offers an on-premises mode using Docker with no account, or a cloud mode with a free key. A connect command wires it to coding agents such as Claude Code, Cursor, Codex, Windsurf, Cline, Goose and Copilot. Memory can be exported as plain Markdown in the Open Knowledge Format, imported from Mem0, Letta, Supermemory, Zep or Hindsight, queried as of a past date, and browsed in a local dashboard.

Key features

Extracts decisions, preferences and facts from conversations with memanto remember --from-conversation
Flags conflicts and supersedes outdated memories instead of appending duplicates
Runs a daily consolidation loop after memanto schedule enable
Exports memory as Markdown in the Open Knowledge Format and imports from Mem0, Letta, Supermemory, Zep and Hindsight
Point-in-time recall with the --as-of and --changed-since options
Local dashboard opened with memanto ui

When to use it

Giving several coding agents one shared memory instead of separate silos
Finding out what a fleet of agents believed on a past date
Moving agent memory between frameworks without a proprietary export

Who it is for: Developers running more than one AI agent who want memory curated and portable rather than just stored.

How it fits with Hermes Agent

The repository carries the hermes-agent topic and is built to give several agents one shared memory; the opening part of the README documents connectors for other coding agents and shows no Hermes-specific steps.

How to install Memanto

These commands are copied from the project's README. Check the repository for the latest steps before you run them.

pip install memanto
npm install @moorcheh-ai/memanto
go get github.com/moorcheh-ai/memanto/sdks/go

Requirements: Docker and Ollama for the on-premises mode (no account), or a free key for the cloud mode; runs on macOS, Linux and Windows

FAQ

What is Memanto?

Memanto is a memory agent that manages the memories of your other agents. It decides what to keep, what conflicts, what expires and what each agent needs to know before it acts.

Does Memanto work with Hermes Agent?

The repository is tagged hermes-agent, but the opening part of the README documents connectors for Claude Code, Cursor, Codex, Windsurf, Cline, Goose and Copilot. Check the project documentation for Hermes-specific setup.

How do I install Memanto?

Run pip install memanto, then run the memanto command and choose the On-Prem or Cloud mode. Packages for npm and Go are also listed in the README.