diff --git a/contrib/tools/parallel-search/README.md b/contrib/tools/parallel-search/README.md
new file mode 100644
index 000000000..6ab777312
--- /dev/null
+++ b/contrib/tools/parallel-search/README.md
@@ -0,0 +1,36 @@
+# Parallel Search MCP
+
+This directory registers the public [Parallel Search MCP](https://docs.parallel.ai/integrations/mcp/search-mcp)
+endpoint in kagent as a `RemoteMCPServer`. The default endpoint requires no
+account or API key.
+
+## What it provides
+
+| Tool | Purpose |
+|------|---------|
+| `web_search` | Searches the live web for current information. |
+| `web_fetch` | Extracts clean Markdown from a URL. |
+
+## Installation
+
+```bash
+kubectl apply -f parallel-search-remote-mcpserver.yaml
+```
+
+This creates a `RemoteMCPServer` named `parallel-search` that points to
+`https://search.parallel.ai/mcp`. Agents must explicitly reference the server
+and select its tools; applying the manifest does not change existing agents.
+
+## Verify
+
+```bash
+kubectl get remotemcpserver parallel-search -n kagent -o yaml
+```
+
+The `Accepted` condition should become true, and `status.discoveredTools`
+should list `web_search` and `web_fetch`.
+
+## Learn more
+
+- [Parallel Search MCP documentation](https://docs.parallel.ai/integrations/mcp/search-mcp)
+- [MCP protocol](https://modelcontextprotocol.io/)
diff --git a/contrib/tools/parallel-search/parallel-search-remote-mcpserver.yaml b/contrib/tools/parallel-search/parallel-search-remote-mcpserver.yaml
new file mode 100644
index 000000000..77e04f727
--- /dev/null
+++ b/contrib/tools/parallel-search/parallel-search-remote-mcpserver.yaml
@@ -0,0 +1,14 @@
+## Parallel Search MCP
+## https://search.parallel.ai/mcp
+## Provides live web search and URL fetching without an account or API key.
+apiVersion: kagent.dev/v1alpha2
+kind: RemoteMCPServer
+metadata:
+  name: parallel-search
+  namespace: kagent
+spec:
+  url: "https://search.parallel.ai/mcp"
+  protocol: STREAMABLE_HTTP
+  timeout: 30s
+  sseReadTimeout: 5m0s
+  description: "Live web search and URL fetching through Parallel Search MCP"