hermes-ecosystem
🗺️ Hermes Atlas — the community map of every tool, skill, and integration for Hermes Agent by Nous Research. Live at hermesatlas.com
Overview
Hermes Atlas is a community-curated directory and ecosystem map for the Hermes Agent by Nous Research. It aggregates over 80 quality-filtered repositories, including skills, plugins, and deployment templates, using a vanilla JavaScript frontend and Vercel serverless functions. The platform features a RAG-powered chatbot that answers technical questions grounded in 27 research files using hybrid retrieval and OpenRouter LLMs. Users can search and filter the ecosystem while viewing live GitHub star counts and growth trends cached via Redis.
- Directory of 80+ security-reviewed Hermes Agent tools and integrations
- RAG chatbot providing grounded answers from 27 research files
- Live GitHub star tracking with Redis-cached sparklines and trends
Full README from GitHub
🗺️ Hermes Atlas
The community-curated map of every tool, skill, and integration for Hermes Agent by Nous Research.
🌐 Live site: hermesatlas.com
What is this?
Hermes Atlas is a living directory of the Hermes Agent ecosystem. Hermes Agent (the self-improving AI agent from Nous Research) launched in February 2026 and immediately spawned a fast-growing community of skills, plugins, integrations, deployment templates, and forks. This site is the canonical map of all of it.
Features:
- 80+ quality-filtered repos across 12 categories — every project security-reviewed before inclusion
- Live star counts fetched from the GitHub API and cached in Redis
- Sparklines and trending badges showing growth over time
- RAG-powered chatbot ("Ask the Atlas") that answers questions about Hermes Agent grounded in 27 research files
- Search, sort, filter across the entire ecosystem
- Light and dark mode with OS preference detection
- Mobile responsive
How it works
┌─────────────────────────────────────────────────────────────┐
│ hermesatlas.com (Vercel static + serverless) │
│ │
│ ┌─────────────┐ ┌──────────────┐ ┌──────────────────┐ │
│ │ index.html │ │ /api/stars │ │ /api/chat │ │
│ │ (the map) │ │ (live data) │ │ (RAG chatbot) │ │
│ └─────────────┘ └──────┬───────┘ └────────┬─────────┘ │
│ │ │ │
│ ▼ ▼ │
│ ┌──────────┐ ┌────────────┐ │
│ │ Redis │ │ OpenRouter │ │
│ │ Cloud │ │ (Gemma 4) │ │
│ └────┬─────┘ └────────────┘ │
│ │ │
│ ▼ │
│ ┌──────────┐ │
│ │ GitHub │ │
│ │ GraphQL │ │
│ └──────────┘ │
└─────────────────────────────────────────────────────────────┘
Stack:
- Frontend: Vanilla HTML/CSS/JavaScript (no framework, no build step)
- Hosting: Vercel (static + serverless functions)
- Cache: Redis Cloud (1hr TTL on star counts, daily history snapshots)
- LLM: OpenRouter with fallback chain (Gemma 4 31B → Gemma 4 26B → Gemini 3 Flash)
- Embeddings: OpenAI text-embedding-3-small (computed once at build time, cached as static JSON)
- Retrieval: Hybrid BM25 + cosine similarity, MMR re-ranking, conversation-aware query rewriting
Repository structure
hermes/
├── index.html # The map (single-page app)
├── api/
│ ├── stars.js # GitHub star fetch + Redis cache
│ ├── stars-history.js # 30-day history for sparklines
│ └── chat.js # RAG pipeline with streaming
├── lib/
│ └── redis.js # Shared Redis client helper
├── data/
│ ├── repos.json # 84 quality-filtered repos (single source of truth)
│ └── chunks.json # Pre-computed text embeddings (~7MB, 283 chunks)
├── scripts/
│ ├── build-chunks.js # Splits research/ into chunks + embeds them
│ └── test-rag.js # Local RAG quality tests (27/27 passing)
├── research/ # 27 source-of-truth research files
├── repos/ # Star count data, security review, raw search results
├── ECOSYSTEM.md # Markdown version of the map
├── package.json # Just two deps: openai, redis
└── vercel.json # Function config + daily cron
Running locally
git clone https://github.com/ksimback/hermes-ecosystem.git
cd hermes-ecosystem
# Install deps (only needed for the API endpoints + chunk builder)
npm install
# To rebuild the chatbot's knowledge base after editing research/ files:
OPENROUTER_API_KEY=sk-or-... node scripts/build-chunks.js
# To test the RAG pipeline locally:
OPENROUTER_API_KEY=sk-or-... node scripts/test-rag.js
# To preview the static site, just open index.html in a browser
# (the API endpoints only work when deployed to Vercel)
Environment variables (Vercel)
| Variable | Purpose | Required |
|---|---|---|
GITHUB_TOKEN |
Fine-grained PAT, public repos read-only | For 5000/hr rate limit (60/hr without) |
OPENROUTER_API_KEY |
LLM API key for chat | Yes |
REDIS_URL |
Redis Cloud connection string | Yes (for cache + history) |
OPENROUTER_MODEL |
Override primary LLM model | No (default: google/gemma-4-31b-it:free) |
OPENROUTER_FALLBACK_MODELS |
Comma-separated fallback chain | No |
Contributing
Found a Hermes Agent project that should be in the map? Open an issue with the GitHub URL. Filtering criteria:
- Must be specifically built for or integrated with Hermes Agent
- Created after July 22, 2025 (Hermes repo creation date)
- Not a personal pet project or assignment
- Shows genuine effort and adds value to the ecosystem
- Passes a basic security review
License
Site code: MIT. Research content: CC BY 4.0. Repo descriptions and metadata are sourced from the upstream projects' own documentation.
Built by @ksimback. Not officially affiliated with Nous Research — community project celebrating their work.