Skip to main content

Next-generation memory provider for Hermes Agent — fusing OpenViking vector storage with MemOS cognitive engine

Project description

Hermes Next

Next-generation memory provider for Hermes Agent.

Hermes Next fuses OpenViking vector storage with a Python-native MemOS cognitive engine, giving Hermes Agent agents persistent,结构化记忆能力。

Features

  • OpenViking Backend — Long-term vector storage, semantic retrieval, session management
  • MemOS Cognitive Pipeline — L1 Trace capture → reward backpropagation → L2 Policy induction → L3 World Model → Skill crystallization
  • Python Native — Zero bridging overhead, runs in-process
  • Local SQLite Cache — FTS5 full-text search + numpy-based cosine similarity, zero dependencies beyond stdlib
  • Fusion Retrieval — 6-step pipeline combining semantic search, full-text search, policy matching, timeline, recency boost, and MMR diversification

Installation

pip install hermes-next

Requires Python 3.10+ and a running OpenViking server (v0.3.22+).

Configuration

Create a hermes-next.yaml in your config directory:

openviking:
  base_url: "http://localhost:1933"
  api_key: null

cache:
  path: "~/.hermes-next/cache.db"
  enable_fts: true

agent:
  name: "default"

Or set environment variables:

export HERMES_NEXT_OV_URL="http://localhost:1933"
export HERMES_NEXT_CACHE_PATH="~/.hermes-next/cache.db"

Usage with Hermes Agent

from hermes_next import HermesNextProvider

provider = HermesNextProvider()
provider.initialize(session_id="my-session")

# The provider handles prefetching, storage, and retrieval automatically
context = provider.prefetch("What did we discuss about RAG?")
print(context)

Or via CLI:

hermes agent --memory-provider hermes-next

Tools

The provider exposes these tools to the agent:

Tool Description
memos_search(query, k) Semantic search across traces
memos_get(trace_id) Read a specific trace
memos_timeline(limit) Recent activity timeline

Project Structure

hermes-next/
├── hermes_next/
│   ├── ov/            # OpenViking REST client
│   ├── memos/         # MemOS cognitive engine
│   ├── cache/         # SQLite local cache
│   └── retrieval/     # Fusion retrieval pipeline
├── tests/
└── plugin.yaml        # Hermes plugin manifest

License

AGPL-3.0 — This project is a derivative of OpenViking (AGPL-3.0).

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

hermes_next-0.3.1.tar.gz (62.3 kB view details)

Uploaded Source

Built Distribution

If you're not sure about the file name format, learn more about wheel file names.

hermes_next-0.3.1-py3-none-any.whl (69.2 kB view details)

Uploaded Python 3

File details

Details for the file hermes_next-0.3.1.tar.gz.

File metadata

  • Download URL: hermes_next-0.3.1.tar.gz
  • Upload date:
  • Size: 62.3 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/6.1.0 CPython/3.13.12

File hashes

Hashes for hermes_next-0.3.1.tar.gz
Algorithm Hash digest
SHA256 1de121600a74feb9f1e2ac33f68f34da932e9719edde2aaefa608b4878e65ce9
MD5 27b56cde51eed5dc0371140c7e273700
BLAKE2b-256 36e9c207f79154c2f1777d85ccd05154eae834969cb9cc161a2e144c1b9e996e

See more details on using hashes here.

File details

Details for the file hermes_next-0.3.1-py3-none-any.whl.

File metadata

  • Download URL: hermes_next-0.3.1-py3-none-any.whl
  • Upload date:
  • Size: 69.2 kB
  • Tags: Python 3
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/6.1.0 CPython/3.13.12

File hashes

Hashes for hermes_next-0.3.1-py3-none-any.whl
Algorithm Hash digest
SHA256 4d3847eb9c6cbfaa54d5ad7896c2d7f322c5796a80f5348b192fc8cd97a01d4a
MD5 0fa8e986d1472d6577f19a9e24074fa3
BLAKE2b-256 12c080118a103936961b30d17593f4e457ed6438028d992132042ad4420f7f1e

See more details on using hashes here.

Supported by

AWS Cloud computing and Security Sponsor Datadog Monitoring Depot Continuous Integration Fastly CDN Google Download Analytics Pingdom Monitoring Sentry Error logging StatusPage Status page