AIUseCaseHub MCP
AIUseCaseHub MCP
Search source-linked enterprise AI deployments from your agent. Find examples by company, industry, technology, cloud provider and reported outcomes, then retrieve the underlying case for evidence and context.
Website · MCP documentation · Get an API key · Official registry record
This repository contains an optional runnable stdio adapter, integration examples and registry metadata for the hosted service. Connecting directly to the existing endpoint requires no server installation, hosting account or npm package.
Connect
Choose a client supporting Streamable HTTP.
Add the endpoint below with no authentication to try the public preview. No registration or key is needed.
For full results, sign in to AIUseCaseHub's API page, generate a personal key and send it as
X-API-Key.
https://basic-backend713-dsd8hygyfkc8d8f5.swedencentral-01.azurewebsites.net/api/mcpPublic preview configuration for clients using an mcpServers map:
{
"mcpServers": {
"aiusecasehub": {
"url": "https://basic-backend713-dsd8hygyfkc8d8f5.swedencentral-01.azurewebsites.net/api/mcp"
}
}
}Optional full-access configuration:
{
"mcpServers": {
"aiusecasehub": {
"url": "https://basic-backend713-dsd8hygyfkc8d8f5.swedencentral-01.azurewebsites.net/api/mcp",
"headers": {
"X-API-Key": "<YOUR_PERSONAL_API_KEY>"
}
}
}
}Use your client's secret storage or environment-variable support for personal keys. Omit the entire authentication header for a free preview; an empty or invalid key remains an authentication error. Configuration formats vary by client.
The service supports the July 2026 stateless protocol and the older initialization handshake over Streamable HTTP (2024-11-05, 2025-03-26, 2025-06-18, 2025-11-25). Current clients send request metadata on each POST; older clients negotiate through initialize. The direct endpoint offers an anonymous preview and optional personal API-key authentication.
Related MCP server: AssetLab MCP Server
Run the stdio adapter
For clients and directory scanners that require a local process, this repository provides a stdio adapter using mcp-remote. It forwards requests to the real hosted MCP and returns its live tool definitions and results. It requires outbound HTTPS access to the endpoint above; it does not include a local database or the hosted backend.
With Node.js 22 or newer:
git clone https://github.com/flo7up/aiusecasehub-mcp.git
cd aiusecasehub-mcp
npm ci --ignore-scripts
node server.mjsThe process waits for MCP JSON-RPC on stdin. In a client configuration, use node as the command and the absolute path to server.mjs as its argument. Use node server.mjs directly rather than npm start in MCP clients, so npm's banner cannot enter the protocol stream.
Alternatively, build and run the container:
docker build -t aiusecasehub-mcp:local .
docker run --rm -i aiusecasehub-mcp:localNo key is required to start, initialize, list tools or use the limited preview. For full access, set AIUSECASEHUB_API_KEY in your client's secret environment; with Docker, add -e AIUSECASEHUB_API_KEY to forward an already-set environment variable. The adapter sends it as X-API-Key. An explicitly empty key fails locally; an invalid key remains an upstream authentication error. Omit the variable entirely for the preview.
The Docker image runs as a non-root user and installs locked dependencies during build. It publishes no HTTP port. The transport bridge is pinned to mcp-remote 0.8.2; its qs dependency is overridden to patched version 6.16.0.
To verify initialization, four live tool definitions and ping without using preview attempts:
node examples/verify-stdio.mjs
node examples/verify-stdio.mjs --dockerAdd --exercise to either command to perform one real search and detail lookup (two anonymous preview attempts). This verifier always omits personal keys. CI builds the container and checks both transports without consuming tool-call credits.
Glama build configuration
Submit this public repository as AIUseCaseHub MCP. glama.json declares flo7up as the maintainer. The checked-in Dockerfile is runnable without credentials. If Glama asks for build configuration in its Dockerfile admin page, use Node.js 24, build command npm ci --omit=dev --ignore-scripts, and command arguments ["node", "server.mjs"] from the repository root. Leave API-key configuration unset for evaluation. All four tools are discovered from the hosted service, not from a static server card.
Free preview and full access
Public preview | Personal API key | |
Registration | Not required | Required |
Search results | Up to 3 | Up to 20 |
Case content | Source/context fields; descriptions up to 360 characters | Full published case data |
Tool-call credits | 0 | 2 per search; 1 per detail lookup |
Preview allowance | 10 tool-call attempts per IP per UTC hour | Does not consume the preview allowance |
The anonymous service has a shared budget of 200 tool-call attempts per UTC hour. The limits are stored centrally and remain effective across workers and restarts. Shared gateways may share an IP allowance. Discovery and tool listing do not consume it. Rate-limit errors include retry guidance and a link to full access; a storage failure closes the preview temporarily instead of allowing unlimited calls.
Four tools
Tool | Use | Required argument | Credits with a personal key |
| General natural-language search with hybrid ranking |
| 2 |
| Hybrid search with company, industry, geography, provider and technology filters |
| 2 |
| Semantic similarity search |
| 2 |
| Retrieve a case returned by search |
| 1 |
Search tools accept limit from 1 to 20, capped at 3 for public previews. Hybrid search also supports industry, country, technologies_used, cloud_provider, customer_name, partner_name, and boolean case-type filters such as isAgentCase and isRag. See the complete tool reference.
Search payloads include items, total, searchMode, dataSnapshot, and _mcp credit metadata. Pass an item's RowKey to get_usecase_details. Parse the MCP text content as JSON; use the source links and evidence fields when interpreting reported outcomes. A deployment example is not a guarantee of the same results elsewhere.
Example prompts:
Find AI deployments in insurance that report measurable claims-processing outcomes. Retrieve the most relevant cases and cite their sources.
Find manufacturing examples using retrieval-augmented generation, and compare their implementation approaches.
Search for a company's AI deployments, retrieve the supporting case records, and separate reported outcomes from missing evidence.
Verify your connection
The included Node.js 20+ script uses built-in APIs and needs no package installation. By default it uses the anonymous preview. For full access it reads a personal key from a local text file or the AIUSECASEHUB_API_KEY environment variable. It never prints the key.
# Free discovery and all four tool definitions.
node examples/verify-connection.mjs
# Free search and abbreviated detail lookup; uses 2 preview attempts.
node examples/verify-connection.mjs --exercise
# Discovery and tool listing only; no tool-call credits are consumed.
node examples/verify-connection.mjs --key-file 'C:\private\aiusecasehub-key.txt'
# One search followed by one detail lookup: 3 credits on successful completion.
node examples/verify-connection.mjs --key-file 'C:\private\aiusecasehub-key.txt' --exerciseThe script verifies all four tool names and stops on HTTP, protocol or tool errors. Its output describes the checks actually completed; it does not imply that a different client or gateway was tested.
Registry metadata
Glama also lists the hosted connector, with a Healthy status and all four tools verified on September 6, 2026.
The service is also listed on Smithery. Its optional configuration collects each user's personal AIUseCaseHub key; leave it unset for the preview. A public-preview search and detail lookup through Smithery's connection API were verified on September 6, 2026.
Smithery CLI 1.2.0 currently defaults to a gateway address that returned 404 during verification. If affected, use Smithery's documented API base for connections in the current PowerShell session:
$env:SMITHERY_CONNECT_BASE_URL = 'https://api.smithery.ai/connect'
smithery mcp add flo7up/aiusecasehub --id aiusecasehub
smithery tool list aiusecasehubSmithery requires its own sign-in for gateway connections. Connecting directly to the hosted endpoint above requires no account for the preview.
The official MCP Registry name is io.github.flo7up/aiusecasehub, version 1.1.1. server.json declares an optional secret header without embedding a credential.
smithery-config.json declares the optional personal API-key header for Smithery's configuration UI. Free previews use no key; full-access users supply their own. Publication and working connections are verified separately.
Credits, data and support
The adapter code and documentation in this repository are MIT licensed. The hosted service, its private implementation and the case dataset are not part of this repository or that license; service access remains subject to the linked terms and preview/credit limits.
With a personal key, successful searches cost 2 credits and detail lookups cost 1. The public preview, discovery and listing do not deduct credits. The response's _mcp object describes preview limits or reports credits used and remaining. Account allowances and purchases are managed on the API page.
Available Tools
4 toolsget_usecase_detailsRead an AI deployment caseARead-only
Fetch detailed information for a single AI use case by RowKey/id.
Inputs: row_key (required), id.
hybrid_search_usecasesSearch AI deployments with filtersARead-only
Search AI use cases with hybrid full-text and vector ranking. Supports provider, industry, geography, technology, customer, partner, and boolean case-type filters.
Inputs: query (required), isRag, limit, country, +14 more.
search_usecasesSearch enterprise AI deploymentsARead-only
Search curated AI use cases using the default hybrid ranking. Use this when an agent has a natural-language search term.
Inputs: search_term (required), limit, query.
vector_search_usecasesFind semantically similar AI deploymentsARead-only
Semantic vector search for AI use cases using the meaning of the query text.
Inputs: query (required), limit.
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
TDQS
Scored across 4 tools
search_usecases and hybrid_search_usecases are nearly the same operation—both perform hybrid full-text/vector ranking—and the only clear distinction is that the latter adds filters. An agent will struggle to know which to choose. vector_search_usecases and get_usecase_details are distinct, but the overlapping search tools create real ambiguity.
All tool names follow a consistent verb_noun snake_case pattern: search_usecases, hybrid_search_usecases, vector_search_usecases, get_usecase_details. The prefixes and suffixes are predictable and readable.
Four tools is a reasonable size for a focused use-case search and detail server. However, one of the four tools is largely redundant with another, so the count is slightly padded rather than perfectly minimal.
The server covers the core workflow of discovering AI use cases via search and then fetching full details. It lacks an explicit browse/list-all or filter-only endpoint, but the search tools likely handle most discovery needs, so the gaps are minor.
Maintenance
Related MCP Connectors
Source-linked enterprise AI deployment search. Free preview; API keys unlock full results.
Search and read the aicoolies catalog of AI and developer tools via a remote MCP knowledge graph.
Find source-linked AI tool recipes and retrieve complete, reproducible run kits.
Patterns for designing and reviewing AI skills, agents, and multi-agent workflows. Find guidance on context economy, delegation, verification, and tool design; inspect claims, worked examples, maturity labels, and source references. Five read-only tools let agents discover relevant patterns, compare concise cards, read specific sections, and explore relationships. Hosted Streamable HTTP at https://agentic-atlas.dev/mcp/ — no installation, account, or API key required. Browse the atlas at https://agentic-atlas.dev/.
Related MCP Servers
- FlicenseAqualityDmaintenanceEnables AI assistants to search blog posts and pages from Ghost, and learn pages, case studies, and events from Contentful. Provides five search tools with contextual snippets via MCP stdio transport.5-
- AlicenseBqualityBmaintenanceConnects Claude, ChatGPT, Microsoft Copilot, and other MCP clients to an AssetLab asset management account, letting users query and manage work orders, PM schedules, capital projects, infrastructure networks, and other records in plain language. It exposes 480+ scoped, tenant-bound, audited tools with both read and write access, hosted or run locally over stdio.481230 npmMIT
- AlicenseNot gradedqualityBmaintenanceEnables MCP clients such as Claude, Claude Code and Cursor to keyword-search a directory of over 1,000 hand-picked productivity tools, apps, AI tools, courses and books, filtering by category, type, pricing and result limit, and to fetch a single listing by slug, page URL or exact name and list all categories with counts. It runs read-only over stdio with no API key, dependencies or tracking, making only a plain GET of the public catalogue.MIT
- AlicenseAqualityBmaintenanceStdio proxy for HirePulse: 27 live tools for company research, hiring, job changes, public contacts and commerce evidence. Focused remote profiles: 9 B2B or 11 commerce tools. Free sandbox, 10 successful calls per agent per UTC day (shared network ceiling), and eligible 25-call trials. Paid results use x402 USDC on Base. Install: npx -y hirepulse-mcp (Node.js 20+).2649 npm1MIT