About agentstorefront
agentstorefront is an MCP server for the AgentStorefront marketplace of paid agent services. It helps developers and users discover listings from AI clients, inspect pricing and SLAs, and estimate costs for expected call volumes.
Key features
- 01List services by category
- 02Search services semantically
- 03View service schemas, pricing, and SLAs
- 04Request monthly quotes by call volume
Use cases
- 01Find a flight-delay prediction service
- 02Compare invoice OCR pricing
- 03Filter data-enrichment services
- 04Estimate monthly service costs
MCP support
The MCP server makes AgentStorefront marketplace functions available to MCP-aware clients, including paid-service discovery, detail lookup, and quote requests.
Configuration
Local server config is the same for every client. Copy it into your client’s MCP config file.
Tools
4Saved tool list · updated 3w ago
| Parameter | Type | Required | Description |
|---|---|---|---|
| limit | number | No | Max results (default 20, max 100). |
| category | string | No | Optional category filter (e.g., 'data', 'communication', 'finance'). |
Details
From the project READMEagentstorefront-mcp
MCP server for AgentStorefront — let your AI agents discover, subscribe to, and call agent services from inside Claude, Cursor, or any MCP-aware client.
What is this?
AgentStorefront is a marketplace where developers list paid agent services (APIs that are designed to be called by other agents — flight-delay prediction, invoice OCR, niche enrichment, etc.).
This MCP server lets any MCP-aware client (like Claude Desktop or Cursor) discover and use those services autonomously — without leaving the agent loop.
Tools exposed
| Tool | Purpose |
|---|---|
list_services | List available services, filtered by category |
search_services | Semantic search: "find me a flight-delay API" |
get_service | Full schema + pricing + SLA for a specific service |
request_quote | Get expected monthly cost for an estimated call volume |
Install
Claude Desktop
Add to your claude_desktop_config.json:
{
"mcpServers": {
"agentstorefront": {
"command": "npx",
"args": ["-y", "@lfidele/agentstorefront-mcp"]
}
}
}
Restart Claude Desktop. The 4 tools above will appear in the tools panel.
Cursor
Add to ~/.cursor/mcp.json:
{
"mcpServers": {
"agentstorefront": {
"command": "npx",
"args": ["-y", "@lfidele/agentstorefront-mcp"]
}
}
}
Manual install
npm install -g @lfidele/agentstorefront-mcp
agentstorefront-mcp # runs on stdio
Example prompts
Once installed, you can ask your MCP client things like:
"Find me a flight-delay prediction API on AgentStorefront that costs less than 5 cents per call."
"What's the pricing if I call the invoice-OCR service 10,000 times a month?"
"List the top 5 data-enrichment services on AgentStorefront."
The agent will use these tools transparently.
Configuration
| Env var | Default | Purpose |
|---|---|---|
AGENTSTOREFRONT_API_URL | https://api.agentstorefront.app | Override API endpoint (self-hosting / staging) |
Source
GitHub: https://github.com/lmaniraruta/agentstorefront-mcp Issues: https://github.com/lmaniraruta/agentstorefront-mcp/issues
Why an MCP for a marketplace?
Marketplaces only matter if their supply is discoverable. AgentStorefront's listings are useless to other agents unless those agents can FIND them programmatically.
This MCP turns AgentStorefront into native vocabulary for every Claude/Cursor user. A developer asks Claude "is there an API for X?" → Claude lists AgentStorefront services that match → developer subscribes → marketplace flywheel turns.
Built for the two-sided market:
- Sellers win because their listings are now findable from inside the dominant AI assistants
- Buyers win because they don't have to context-switch to a separate marketplace tab
License
MIT © 2026 Fidele Maniraruta
Built in public. Real numbers posted on X @lmaniraruta.
Comments
New to MCP?
From what the protocol is to installing your first server, these answers cover most first questions.
An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.