Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type @ followed by the MCP server name and your instructions, e.g., "@hermes-atlas-mcp Search Hermes Atlas for memory providers"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
MCP server exposing the Hermes Atlas ecosystem catalog — 100+ community-built tools, skills, plugins, memory providers, workspaces, and integrations for Nous Research's Hermes Agent — to any MCP-aware client (Claude Desktop, Cursor, Continue, Windsurf, etc.).
Ask your AI assistant things like:
"Search Hermes Atlas for memory providers"
"What does the hermes-atlas catalog say about the hermes-agent project?"
"List the top skills for Hermes Agent"
"Get the Hermes beginner's guide"
"Ask the atlas: what's the difference between Hermes Agent and Claude Code?"
All catalog data is fetched live from hermesatlas.com (cached in-memory for 1 hour), so tool answers reflect the current ecosystem, not a stale snapshot.
Status: 0.x, experimental. Breaking changes may happen before 1.0. Batch releases weekly.
npx -y hermes-atlas-mcp install --client claude-desktopReplace claude-desktop with cursor or claude-code as needed. The installer finds the right config file for your OS, merges in the hermes-atlas entry without touching your other MCP servers, and tells you what to do next.
Other flags:
--print — dry-run (shows the diff, writes nothing)
--force — overwrite an existing hermes-atlas entry without prompting
To remove:
npx -y hermes-atlas-mcp uninstall --client claude-desktopFully quit and reopen the client after either command (closing the window isn't enough — the process needs to restart).
Find the config file:
Easiest: open Claude Desktop, go to Settings → Developer → Edit Config. The file opens in your default editor.
macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
Windows: %APPDATA%\Claude\claude_desktop_config.json
Linux: ~/.config/Claude/claude_desktop_config.json
Paste this (if the file is empty or doesn't exist) or merge the hermes-atlas entry into your existing mcpServers object:
{
"mcpServers": {
"hermes-atlas": {
"command": "npx",
"args": ["-y", "hermes-atlas-mcp"]
}
}
}Save. Fully quit Claude Desktop and reopen. The five tools appear in the tool picker.
File: ~/.cursor/mcp.json (global) or .cursor/mcp.json (per-project). Same JSON snippet as above.
Either use the installer above, or run /mcp inside Claude Code and add through its UI, or hand-edit ~/.claude.json.
git clone https://github.com/ksimback/hermes-atlas-mcp.git
cd hermes-atlas-mcp
npm install
npm run build
node dist/index.js # stdio server, pipe JSON-RPC to testRelated MCP server: agentalmanac-mcp
Tool | What it does |
| Ranked keyword search over the catalog. Optional |
| Full summary + metadata + URL for one |
| Ranked projects in one of the six curated lists. |
| Full text of the beginner's guide, install guide, or vs-Claude-Code comparison. |
| Free-form RAG question routed to the site's |
URI | Content |
| Full catalog JSON |
| AI-generated summaries JSON |
| Curated lists JSON |
|
|
npm install
npm run dev # tsc --watch
npm run build # tsc
npm start # runs dist/index.js on stdioMIT © Kevin Simback. Catalog data is CC0.
hermesatlas.com — the live site
ksimback/hermes-ecosystem — the source repo behind the site
Hermes Agent — the agent this ecosystem is built around
Ask a free-form natural-language question about the Hermes Agent ecosystem. Delegates to the Hermes Atlas RAG endpoint which searches across repos, summaries, guides, and the ECOSYSTEM.md overview, then returns a cited answer. Good for questions like 'What's the best memory provider for production?' or 'How does Hermes compare to Claude Code?'.
Inputs: question (required).
Fetch the full text of a Hermes Handbook guide page. Args: slug — one of 'hub' (beginner's guide at /guide/), 'install' (/guide/install/), or 'vs-claude-code' (/guide/vs-claude-code/). Any other slug returns an error.
Inputs: slug (required).
Get the full summary, metadata, and canonical URL for a specific Hermes ecosystem project. Args: owner (GitHub owner, e.g. 'NousResearch'), repo (GitHub repo name, e.g. 'hermes-agent').
Inputs: repo (required), owner (required).
Return the ranked list of projects in a Hermes Atlas curated list. Valid slug values: 'best-memory-providers' | 'top-skills' | 'deployment-options' | 'multi-agent-frameworks' | 'developer-tools' | 'workspaces-and-guis'.
Inputs: slug (required).
Search the Hermes Atlas catalog of 100+ community-built Hermes Agent tools, skills, plugins, memory providers, workspaces, and integrations. Query can be natural language ('memory providers', 'telegram bot', 'vscode plugin'). Optional category filter: 'Core & Official' | 'Workspaces & GUIs' | 'Memory & Context' | 'Skills & Skill Registries' | 'Plugins & Extensions' | 'Integrations & Bridges' | 'Multi-Agent & Orchestration' | 'Developer Tools' | 'Deployment & Infra' | 'Domain Applications' | 'Guides & Docs' | 'Forks & Derivatives'. Optional limit (1-50, default 10).
Inputs: query (required), limit, category.
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
Scored across 5 tools
Tools are mostly distinct: search_projects returns a list of projects, get_project returns one project, list_by_category returns from a curated list, get_guide fetches a guide page, and ask_atlas gives a cited answer. The only potential confusion is between search_projects and ask_atlas, but their descriptions clarify the different output types.
All tool names follow a consistent verb_noun pattern in lowercase snake_case (search_, get_, list_, get_, ask_). This is a perfectly uniform naming scheme with no deviations.
Five tools is an appropriate, focused scope for a catalog/atlas server. Each tool covers a distinct need: search, specific lookup, category browsing, guide retrieval, and open-ended Q&A. No bloat or unnecessary duplication.
For a read-only catalog, the surface covers search, detailed retrieval, curated lists, guides, and free-form questions. Minor gaps exist: no tool to list all projects in the catalog or fetch the full ECOSYSTEM.md, and the guide access is limited to three fixed slugs, but these are workable limitations.
Search and inspect source-pinned agent skills, plugins and tools through a free public MCP.
Search and read the aicoolies catalog of AI and developer tools via a remote MCP knowledge graph.
Free, source-checked directory of AI agent tools, MCP servers and skills, with read-only search.
Search a curated directory of agent tools, MCP servers, APIs, protocols and skills. No auth.