diff --git a/contributing/samples/managed_agent/custom_agent/README.md b/contributing/samples/managed_agent/custom_agent/README.md
new file mode 100644
index 00000000..4e0f11ac
--- /dev/null
+++ b/contributing/samples/managed_agent/custom_agent/README.md
@@ -0,0 +1,87 @@
+# Managed Agent: Create and Use a Custom Agent
+
+> For setup, authentication, backends, and background on `ManagedAgent`, see the
+> [ManagedAgent guide](../../../../docs/guides/agents/managed_agent/index.md).
+
+## Overview
+
+This sample demonstrates the **control-plane lifecycle** of a custom managed
+agent: creating a persistent, named agent *resource* — its persona and
+server-side tools baked in — then driving it and deleting it.
+
+You do **not** need a custom resource just to set a persona or server-side
+tools. `ManagedAgent` accepts both inline: `instruction=...` for a persona (see
+the [`system_instruction`](../system_instruction) sample) and
+`tools=[google_search]` for server-side tools (see the [`basic`](../basic)
+sample). Create a custom resource when you instead want a reusable,
+server-managed agent that other apps and sessions can share by id.
+
+This module drives that lifecycle: run it with `--create` to provision the
+resource (reusing the genai client `ManagedAgent` already holds,
+`root_agent.api_client`, which exposes both interactions and agent
+create/delete), then drive `root_agent` with `adk web` / `adk run`, and
+`--delete` to remove it.
+
+## Setup
+
+Custom-agent creation requires the **GEAP / Vertex** backend (`global`
+location); the Gemini API backend cannot create agent resources. For backend
+selection, authentication, and credentials, see the
+[ManagedAgent guide](../../../../docs/guides/agents/managed_agent/index.md#prerequisites).
+
+## Usage
+
+```bash
+# 1. Create the custom agent (once).
+python contributing/samples/managed_agent/custom_agent/agent.py --create
+
+# 2. Chat with it. Provisioning can take a few minutes (longer for the first
+#    agent in a project), so wait a moment after --create before the first turn.
+adk run contributing/samples/managed_agent/custom_agent
+#    or: adk web
+
+# 3. Delete it when done.
+python contributing/samples/managed_agent/custom_agent/agent.py --delete
+```
+
+Creation is asynchronous: `--create` returns before the agent is fully ready, so
+if the first turn fails with a "not found" / "being created" error, wait a few
+seconds and retry.
+
+## Sample Inputs
+
+Answers are grounded in live search, so exact text varies:
+
+- `What are the most significant AI announcements this week?`
+
+  The created agent's persona makes it answer **concisely** and **cite its
+  sources**, using server-side `google_search`.
+
+- `Summarize that in one sentence.`
+
+  A follow-up turn that reuses the recovered interaction (multi-turn chaining).
+
+## Graph
+
+```mermaid
+graph LR
+    User -->|message| CustomManagedAgent
+    CustomManagedAgent -->|interactions.create| ManagedAgentsAPI
+    ManagedAgentsAPI -->|server-side google_search| ManagedAgentsAPI
+    ManagedAgentsAPI -->|streamed events| CustomManagedAgent
+    CustomManagedAgent -->|answer| User
+```
+
+## How To
+
+- **Define the custom agent**: pass a `system_instruction` (persona) and
+  server-side `tools` (here `{'type': 'google_search'}`) to
+  `client.agents.create(...)`, extending the `antigravity-preview-05-2026` base
+  agent.
+- **Reuse the ManagedAgent client**: `root_agent.api_client` is the genai client
+  `ManagedAgent` already holds; its `agents.create` / `agents.delete` cover the
+  control plane.
+- **Provision a sandbox**: `ManagedAgent(environment={'type': 'remote'})` gives
+  each interaction a remote sandbox (required to run the agent).
+- **Run it**: `--create` provisions, `--delete` removes; in between, `root_agent`
+  is a normal `BaseAgent`, so `adk web` / `adk run` (or a `Runner`) drive it.
diff --git a/contributing/samples/managed_agent/custom_agent/__init__.py b/contributing/samples/managed_agent/custom_agent/__init__.py
new file mode 100644
index 00000000..4015e47d
--- /dev/null
+++ b/contributing/samples/managed_agent/custom_agent/__init__.py
@@ -0,0 +1,15 @@
+# Copyright 2026 Google LLC
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+from . import agent
diff --git a/contributing/samples/managed_agent/custom_agent/agent.py b/contributing/samples/managed_agent/custom_agent/agent.py
new file mode 100644
index 00000000..fbea1216
--- /dev/null
+++ b/contributing/samples/managed_agent/custom_agent/agent.py
@@ -0,0 +1,90 @@
+# Copyright 2026 Google LLC
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+"""Create, use, and delete a custom managed-agent resource.
+
+This sample demonstrates the control-plane lifecycle of a custom managed agent:
+creating a persistent, named agent *resource* (its persona and server-side tools
+baked in), then driving it and deleting it.
+
+You don't need a custom resource just to set a persona or server-side tools --
+``ManagedAgent`` accepts both inline (``instruction=...`` and
+``tools=[google_search]``; see the ``system_instruction`` and ``basic``
+samples). Create a custom resource when you instead want a reusable,
+server-managed agent that other apps and sessions can share by id.
+
+Run this module with ``--create`` once to provision the resource, then drive
+``root_agent`` with ``adk web`` / ``adk run
+contributing/samples/managed_agent/custom_agent``, then ``--delete`` to remove
+it. See the README for the required GEAP/Vertex setup.
+
+    python contributing/samples/managed_agent/custom_agent/agent.py --create
+    python contributing/samples/managed_agent/custom_agent/agent.py --delete
+"""
+
+import argparse
+
+from dotenv import load_dotenv
+from google.adk.agents import ManagedAgent
+
+load_dotenv()
+
+_AGENT_ID = 'adk-custom-search-agent'
+
+_SYSTEM_INSTRUCTION = (
+    'You are a concise research assistant. Use Google Search to ground every '
+    'answer in current sources, cite the sources you used, and keep answers to '
+    'a few sentences.'
+)
+
+root_agent = ManagedAgent(
+    name='custom_managed_agent',
+    agent_id=_AGENT_ID,
+    environment={'type': 'remote'},
+)
+
+
+def main() -> None:
+  """Create or delete the custom managed-agent resource."""
+  parser = argparse.ArgumentParser(
+      description='Create or delete the custom managed agent for this sample.'
+  )
+  parser.add_argument(
+      '--create', action='store_true', help='Create the custom managed agent.'
+  )
+  parser.add_argument(
+      '--delete', action='store_true', help='Delete the custom managed agent.'
+  )
+  args = parser.parse_args()
+  if not (args.create or args.delete):
+    parser.print_help()
+    return
+
+  # ManagedAgent's genai client also exposes agent create/delete.
+  client = root_agent.api_client
+  if args.create:
+    client.agents.create(
+        id=_AGENT_ID,
+        base_agent='antigravity-preview-05-2026',
+        system_instruction=_SYSTEM_INSTRUCTION,
+        tools=[{'type': 'google_search'}],
+    )
+    print(f'Created "{_AGENT_ID}".')
+  if args.delete:
+    client.agents.delete(id=_AGENT_ID)
+    print(f'Deleted "{_AGENT_ID}".')
+
+
+if __name__ == '__main__':
+  main()