io.github.deficlowPublished by GitHub account deficlowv0.1.1Updated Oct 24, 2026
Search and inspect 6,500+ curated AI apps from the HyperStore directory.
The publisher's own server: Zipp
Exposes the HyperStore directory of 6,500+ curated AI apps so an assistant can search, filter and compare tools by keyword, intent, category, audience or use…
Web executable via Streamable HTTP. Remote servers run from the web runtime once configured in a workspace.
Paste this prompt into Claude Code, Codex, Cursor or another agent — or into SourceWeft chat. The agent reads this server's install guide, shows you its publisher, endpoint and sign-in, and adds it to your client once you agree. Keys never go into a command.
Read https://sourceweft.com/mcp/io.github.deficlow%2Fhyperstore-mcp/install.md and connect the MCP server it describes to my client. Before changing any config, show me its publisher, endpoint and sign-in method and wait for my OK. Never put an API key in a command or in a file I would commit; ask me where it should come from.Add this to your client's mcpServers config.
{
"mcpServers": {
"hyperstore": {
"type": "http",
"url": "https://mcp.store.hypergpt.ai/mcp"
}
}
}Plug 6,500+ AI apps into any LLM via the Model Context Protocol.
[PyPI] [Glama] [Smithery] [MCP Registry] [CI] [License: MIT]
HyperStore is a curated directory of 6,500+ AI applications, developed by HyperGPT. This MCP server exposes the HyperStore catalog to any LLM client — Claude, ChatGPT, Cursor, Windsurf, Cline, Zed, Gemini, and anything else that speaks MCP.
Ask your LLM:
"Find me a free AI tool that summarises PDFs." "Compare ChatGPT, Claude, and Gemini side-by-side." "Show me the top 5 image-generation apps with an API."
The LLM calls HyperStore MCP behind the scenes and answers with up-to-date, curated results.
13 tools:
3 resources:
hyperstore://app/{slug} — markdown rendering of any apphyperstore://category/{slug} — top apps in a categoryhyperstore://catalog — full category index3 prompts:
find_tool_for_task — guided discovery for a taskcompare_apps — side-by-side app comparisondiscover_category — explore a topicuvx (zero install, recommended)Requires uv. One command and you're done:
pipxUse our managed Streamable HTTP server:
Edit ~/Library/Application Support/Claude/claude_desktop_config.json
(macOS) or %APPDATA%\Claude\claude_desktop_config.json (Windows):
Restart Claude → tools appear in the 🛠 menu.
.cursor/mcp.json (project) or ~/.cursor/mcp.json (global):
~/.codeium/windsurf/mcp_config.json:
settings.json:
~/.config/zed/settings.json:
~/.gemini/settings.json:
Settings → Connectors → Add custom connector:
https://mcp.store.hypergpt.ai/mcpSee examples/ for ready-to-paste configs for every supported client.
For self-hosting, use the Docker image.
For direct invocation without Docker, the CLI accepts --transport http|sse
(see hyperstore-mcp --help).
When self-hosting, these environment variables can be set
(see .env.example for the full list):
Inspect the running server with the official MCP Inspector:
HyperStore MCP is a thin async wrapper around the HyperStore public REST API. It is read-only — no credentials, no writes, no PII. The same data that powers the website powers the MCP server. Updates land in your LLM the moment they land on the site.
MIT © HyperGPT
Source: README.md at commit 0b66aad
Search HyperStore's AI apps directory by keyword. Returns a paginated list of matching apps with name, slug, short description, pricing, and rating. Use this when the user gives concrete keywords (e.g. 'image upscaler', 'code copilot').
limitqueryrequiredcursorNatural-language semantic search powered by embeddings. Best for fuzzy intent ('a tool that helps me write Python tests', 'something like Midjourney but free'). Returns up to 12 apps ranked by semantic similarity.
promptrequiredFetch the full detail page for a single AI app by slug: long description, features, screenshots, categories, pricing, rating, website URL, source attribution.
slugrequiredPaginated apps listing with optional filters. Combine `category`, `pricing`, and a free-text `query` to drill down. Returns apps sorted by popularity. Use `cursor` (last app id from previous page) to paginate.
limitquerycursorpricingcategoryList all HyperStore categories with app counts. Use this first when the user asks 'what kinds of AI tools are there?' or to discover available category slugs.
Takes no parameters.
Get apps within a specific category. Returns the category metadata plus a paginated list of apps in that category, sorted by popularity.
slugrequiredlimitcursorpricingBrowse apps A-Z by starting letter. Use letter='#' for apps starting with digits or symbols. Useful for alphabetical discovery rather than search.
limitcursorletterrequiredpricingFetch the HyperStore homepage payload: top categories with their featured apps, the trending apps strip, and totals. Good first call to give the user a broad overview.
pageGet curated alternatives to a specific AI app by slug. Returns the app plus a list of competing/similar tools ranked by match confidence. Use this when the user asks 'what are alternatives to X', 'something like X', or 'X vs others'.
slugrequiredlimitList the audience segments HyperStore curates tools for (e.g. 'developers', 'lawyers', 'students'), each with a slug and app count. Call this first to discover audience slugs for `apps_for_audience`.
Takes no parameters.
Get the best AI tools for a specific audience by slug (from `list_audiences`), ranked by relevance then popularity. Use when the user asks 'best AI tools for {role/profession}'. Paginate with `cursor` (last app id from the previous page).
slugrequiredlimitcursorList the use-case taxonomies HyperStore curates tools for (e.g. 'legal-contracts', 'tiktok-shorts'), each with a slug and app count. Call this first to discover use-case slugs for `apps_for_use_case`.
Takes no parameters.
Get AI tools for a specific use case by slug (from `list_use_cases`), ranked by relevance then popularity. Use when the user asks 'AI tools for {task}'. Paginate with `cursor` (last app id from the previous page).
slugrequiredlimitcursor100% of the applicable checks met
How complete this listing is, computed from the registry entry, the endpoint and public sources. Not a rating of the server and not a statement that it is safe or working today: reachability is a dated check, and no installs or reviews go into the score.
uvx hyperstore-mcppipx install hyperstore-mcphyperstore-mcpdocker run --rm -p 8080:8080 ghcr.io/deficlow/hyperstore-mcp# Now MCP Streamable HTTP at http://localhost:8080/mcphttps://mcp.store.hypergpt.ai/mcp{ "mcpServers": { "hyperstore": { "command": "uvx", "args": ["hyperstore-mcp"] } }}claude mcp add hyperstore -- uvx hyperstore-mcp{ "mcpServers": { "hyperstore": { "command": "uvx", "args": ["hyperstore-mcp"] } }}{ "mcpServers": { "hyperstore": { "command": "uvx", "args": ["hyperstore-mcp"] } }}{ "cline.mcpServers": { "hyperstore": { "command": "uvx", "args": ["hyperstore-mcp"] } }}{ "context_servers": { "hyperstore": { "command": { "path": "uvx", "args": ["hyperstore-mcp"] } } }}{ "mcpServers": { "hyperstore": { "command": "uvx", "args": ["hyperstore-mcp"] } }}from openai import OpenAI
client = OpenAI()response = client.responses.create( model="gpt-4.1", tools=[{ "type": "mcp", "server_label": "hyperstore", "server_url": "https://mcp.store.hypergpt.ai/mcp", "require_approval": "never", }], input="Find me 3 free AI tools for writing unit tests.",)print(response.output_text)from anthropic import Anthropic
client = Anthropic()response = client.messages.create( model="claude-opus-4-7", max_tokens=1024, mcp_servers=[{ "type": "url", "url": "https://mcp.store.hypergpt.ai/mcp", "name": "hyperstore", }], messages=[{"role": "user", "content": "Top 5 AI image generators?"}],)git clone https://github.com/deficlow/HyperStore-MCPcd HyperStore-MCPuv sync --all-extrasuv run pytestuv run hyperstore-mcp # stdio mode for local testingnpx @modelcontextprotocol/inspector uvx hyperstore-mcpLLM client ──MCP──▶ hyperstore-mcp ──HTTPS──▶ store.hypergpt.ai/api