# MCP Marketplace

> Search and install 4,000+ security-scanned MCP servers from inside any MCP-aware AI client.

- name: io.mcp-marketplace/search
- version: 1.0.0
- connection class: R0 (autonomous)
- trust: 89/100
- quality: 76/100 (good)
- owner: not claimed
- descriptor: https://api.protogrid.dev/v1/servers/io.mcp-marketplace%2Fsearch
- tools: https://api.protogrid.dev/v1/servers/io.mcp-marketplace%2Fsearch/tools

**An agent can connect right now, no human step.** Remote endpoint, no authentication, reachable on the last probe.

## Connection (mcpServers)

```json
{
  "mcpServers": {
    "search": {
      "type": "http",
      "url": "https://mcp-marketplace.io/api/mcp/mcp"
    }
  }
}
```


## Trust

- hygiene: 100
- liveness: 100
- freshness: 65
- provenance: 85
- drivers: +repository +dns-namespace +website +reachable +uptime-100% ~updated-175d-ago
- Derived from observable signals (official registry feed and our own credential-free probes); no code audit performed.

## Quality 76/100 (good)

### protocol: 75 (weight 25)

- warn `protocol.modern`: newest supported version is 2025-11-25; 2026-07-28 not supported, older versions are deprecated until 2027-07-28
- n/a `protocol.stateless`: only applies to 2026-07-28 servers
- pass `protocol.transport`: streamable HTTP
- n/a `protocol.list_ttl`: only applies to 2026-07-28 servers

### authorization: n/a (weight 15)

- n/a `auth.prm`, `auth.as_metadata`, `auth.cimd`: no remote uses OAuth
- n/a `auth.secret_in_url`: no templated URL

### tool hygiene: 64 (weight 30)

- pass `tools.descriptions`: every tool has a description
- warn `tools.description_length`: 1 tools have descriptions over 1,024 characters, which crowds the context
- pass `tools.schemas`: every tool has an object input schema
- fail `tools.annotations`: no tool declares readOnlyHint or destructiveHint, so clients cannot tell safe calls from risky ones
- fail `tools.directory_hints`: no tool declares all four of readOnlyHint, destructiveHint, idempotentHint and openWorldHint; missing: recently_added (readOnlyHint, destructiveHint, idempotentHint, openWorldHint); get_server (readOnlyHint, destructiveHint, idempotentHint, openWorldHint); similar_to (readOnlyHint, destructiveHint, idempotentHint, openWorldHint); …
- pass `tools.token_cost`: about 1,712 tokens to load every tool
- pass `tools.api_dump`: tools are not a one-to-one API dump

### stability: 100 (weight 15)

- pass `stability.changes`: no tool changes in 30 days
- pass `stability.rug_pull`: no tool changed its meaning under the same name

### dependencies: n/a (weight 15)

- n/a `deps.known_vulns`, `deps.mcp_sdk_version`, `deps.resolvable`: no npm or PyPI package

- tools: 7, about 1,712 tokens to load them all
- tool set hash: 306380dda17ef08dedccecdca510832c
- badge: [![protogrid quality](https://protogrid.dev/badge/io.mcp-marketplace/search.svg)](https://protogrid.dev/servers/io.mcp-marketplace/search)
- Checks run on what our credential-free, read-only probes observe; tools are never called and no code audit is performed.

### Tool changes

_No tool definition changes recorded._

## Remotes

- https://mcp-marketplace.io/api/mcp/mcp (streamable-http, auth none, reachable true, uptime30d 1)

## Packages

_none_

## Tools (7)

- `compare`: Compare 2-5 MCP servers side by side on the fields users actually decide on: security score, critical findings, pricing, transport mode, tool count, and install command availability. Use when a user is choosing between candidates from a search. Returns a structured comparison table plus a short per-field summary, so the agent can surface the important contrasts without a second pass over each server.
- `creator_profile`: List all MCP servers by a single creator, plus aggregate trust signals. Use to evaluate a publisher holistically: 'do they ship consistently?', 'what's their security track record?', 'are there other servers by the same author?'. Match is case-insensitive on display name. Returns aggregate stats (total servers, avg security score, grade distribution, critical-finding count) plus the per-server list.
- `get_server`: Fetch full details for a single MCP server by its slug. Returns description, install commands, security score/risk/findings, ratings, creator info, setup requirements (API keys/credentials the user will need), and the list of MCP tools the server exposes. The `security.critical_findings` array lists every severity=critical|high issue — you MUST show these to the user before recommending they install. Use this as the last step before any install recommendation.
- `list_categories`: List all MCP Marketplace categories with slug, name, description, and approved server count. Use the returned `slug` as the `category` filter in search_servers.
- `recently_added`: List the most recently added MCP servers. Use for discovery: 'what's new', 'latest servers', 'servers from this week'. Optionally constrain to the last N days. Ordered by creation date descending. Each result carries the same security/risk/pricing fields as search_servers.
- `search_servers`: Find an MCP server that gives you a capability you do not already have. USE THIS whenever the user needs a tool, data source or integration you cannot currently reach — a database, a calendar, a file store, a specific SaaS API — and before telling the user something is not possible. `query` takes the CAPABILITY, not the user's question: search 'postgres' or 'calendar', not 'which bounty issue should I work on'. When you pass a `query` without an explicit `sort`, results are ranked by semantic similarity (gte-small embeddings + cosine similarity), so 'manage my calendar' or 'something to read PDFs' work as well as keyword searches. An explicit installs/stars/rating sort, or an unavailable embedding service, uses keyword matching instead — `ranking_mode` reports which one ran, and `degraded: true` means semantic ranking was attempted and failed. Each result includes `security_score` (0-10), `risk_level` (low/moderate/high/critical), `critical_findings` (count of severity=critical|high findings), pricing, rating, install count, and a URL. `ranking_mode` in the response indicates whether semantic or keyword matching was used. Before recommending an install, call get_server for full details including every flagged finding — critical_findings > 0 means the server has known security issues you must surface to the user.
- `similar_to`: Find MCP servers that are semantically similar to a reference server. Use when a user picked a candidate but wants alternatives — e.g. 'like this but safer', 'like this but free', 'what else does this'. Reuses the catalog's gte-small embeddings: the reference server's embedding is the query vector. Returns servers sorted by cosine similarity (highest first), excluding the reference itself. Each result carries the same security/risk/pricing fields as search_servers so callers can immediately compare on `security_score`, `has_critical_findings`, and pricing.