diff --git a/examples/learning.ipynb b/examples/learning.ipynb
new file mode 100644
index 000000000..62313b61e
--- /dev/null
+++ b/examples/learning.ipynb
@@ -0,0 +1,648 @@
+{
+ "cells": [
+  {
+   "cell_type": "markdown",
+   "id": "51466c8d-8ce4-4b3d-be4e-18fdbeda5f53",
+   "metadata": {},
+   "source": [
+    "# Get/Update State\n",
+    "\n",
+    "When running LangGraph agents, you can easily save good threads and use them in the future.\n",
+    "\n",
+    "**Note:** this requires passing in a checkpointer."
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "7cbd446a-808f-4394-be92-d45ab818953c",
+   "metadata": {},
+   "source": [
+    "## Setup\n",
+    "\n",
+    "First we need to install the packages required"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 1,
+   "id": "af4ce0ba-7596-4e5f-8bf8-0b0bd6e62833",
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "\n",
+      "\u001b[1m[\u001b[0m\u001b[34;49mnotice\u001b[0m\u001b[1;39;49m]\u001b[0m\u001b[39;49m A new release of pip is available: \u001b[0m\u001b[31;49m23.3.1\u001b[0m\u001b[39;49m -> \u001b[0m\u001b[32;49m23.3.2\u001b[0m\n",
+      "\u001b[1m[\u001b[0m\u001b[34;49mnotice\u001b[0m\u001b[1;39;49m]\u001b[0m\u001b[39;49m To update, run: \u001b[0m\u001b[32;49mpip install --upgrade pip\u001b[0m\n"
+     ]
+    }
+   ],
+   "source": [
+    "!pip install --quiet -U langchain langchain_openai tavily-python"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "0abe11f4-62ed-4dc4-8875-3db21e260d1d",
+   "metadata": {},
+   "source": [
+    "Next, we need to set API keys for OpenAI (the LLM we will use) and Tavily (the search tool we will use)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 2,
+   "id": "c903a1cf-2977-4e2d-ad7d-8b3946821d89",
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "OpenAI API Key: ········\n",
+      "Tavily API Key: ········\n"
+     ]
+    }
+   ],
+   "source": [
+    "import os\n",
+    "import getpass\n",
+    "\n",
+    "os.environ[\"OPENAI_API_KEY\"] = getpass.getpass(\"OpenAI API Key:\")\n",
+    "os.environ[\"TAVILY_API_KEY\"] = getpass.getpass(\"Tavily API Key:\")"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "f0ed46a8-effe-4596-b0e1-a6a29ee16f5c",
+   "metadata": {},
+   "source": [
+    "Optionally, we can set API key for [LangSmith tracing](https://smith.langchain.com/), which will give us best-in-class observability."
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "95e25aec-7c9f-4a63-b143-225d0e9a79c3",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "os.environ[\"LANGCHAIN_TRACING_V2\"] = \"true\"\n",
+    "os.environ[\"LANGCHAIN_API_KEY\"] = getpass.getpass(\"LangSmith API Key:\")"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "21ac643b-cb06-4724-a80c-2862ba4773f1",
+   "metadata": {},
+   "source": [
+    "## Set up the tools\n",
+    "\n",
+    "We will first define the tools we want to use.\n",
+    "For this simple example, we will use a built-in search tool via Tavily.\n",
+    "However, it is really easy to create your own tools - see documentation [here](https://python.langchain.com/docs/modules/agents/tools/custom_tools) on how to do that.\n"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 1,
+   "id": "d7ef57dd-5d6e-4ad3-9377-a92201c1310e",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "from langchain_community.tools.tavily_search import TavilySearchResults\n",
+    "\n",
+    "tools = [TavilySearchResults(max_results=1)]"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "01885785-b71a-44d1-b1d6-7b5b14d53b58",
+   "metadata": {},
+   "source": [
+    "We can now wrap these tools in a simple ToolNode.\n",
+    "This is a prebuilt node that extracts tool calls from the most recent AIMessage, executes them, and returns a ToolMessage with the results.\n"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 2,
+   "id": "5cf3331e-ccb3-41c8-aeb9-a840a94d41e7",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "from langgraph.prebuilt import ToolNode\n",
+    "\n",
+    "tool_node = ToolNode(tools)"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "5497ed70-fce3-47f1-9cad-46f912bad6a5",
+   "metadata": {},
+   "source": [
+    "## Set up the model\n",
+    "\n",
+    "Now we need to load the chat model we want to use.\n",
+    "Importantly, this should satisfy two criteria:\n",
+    "\n",
+    "1. It should work with messages. We will represent all agent state in the form of messages, so it needs to be able to work well with them.\n",
+    "2. It should work with OpenAI function calling. This means it should either be an OpenAI model or a model that exposes a similar interface.\n",
+    "\n",
+    "Note: these model requirements are not requirements for using LangGraph - they are just requirements for this one example."
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 3,
+   "id": "892b54b9-75f0-4804-9ed0-88b5e5532989",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "from langchain_openai import ChatOpenAI\n",
+    "\n",
+    "model = ChatOpenAI(temperature=0)"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "a77995c0-bae2-4cee-a036-8688a90f05b9",
+   "metadata": {},
+   "source": [
+    "\n",
+    "After we've done this, we should make sure the model knows that it has these tools available to call.\n",
+    "We can do this using the `.bind_tools()` method, common to many of LangChain's chat models.\n"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 4,
+   "id": "cd3cbae5-d92c-4559-a4aa-44721b80d107",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "model = model.bind_tools(tools)"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "e03c5094-9297-4d19-a04e-3eedc75cefb4",
+   "metadata": {},
+   "source": [
+    "## Define the nodes\n",
+    "\n",
+    "We now need to define a few different nodes in our graph.\n",
+    "In `langgraph`, a node can be either a function or a [runnable](https://python.langchain.com/docs/expression_language/).\n",
+    "There are two main nodes we need for this:\n",
+    "\n",
+    "1. The agent: responsible for deciding what (if any) actions to take.\n",
+    "2. A function to invoke tools: if the agent decides to take an action, this node will then execute that action.\n",
+    "\n",
+    "We will also need to define some edges.\n",
+    "Some of these edges may be conditional.\n",
+    "The reason they are conditional is that based on the output of a node, one of several paths may be taken.\n",
+    "The path that is taken is not known until that node is run (the LLM decides).\n",
+    "\n",
+    "1. Conditional Edge: after the agent is called, we should either:\n",
+    "   a. If the agent said to take an action, then the function to invoke tools should be called\n",
+    "   b. If the agent said that it was finished, then it should finish\n",
+    "2. Normal Edge: after the tools are invoked, it should always go back to the agent to decide what to do next\n",
+    "\n",
+    "Let's define the nodes, as well as a function to decide how what conditional edge to take."
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 77,
+   "id": "3b541bb9-900c-40d0-964d-7b5dfee30667",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "# Define the function that determines whether to continue or not\n",
+    "def should_continue(state):\n",
+    "    last_message = state['messages'][-1]\n",
+    "    # If there is no function call, then we finish\n",
+    "    if not last_message.tool_calls:\n",
+    "        return \"end\"\n",
+    "    # Otherwise if there is, we continue\n",
+    "    else:\n",
+    "        return \"continue\""
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "ffd6e892-946c-4899-8cc0-7c9291c1f73b",
+   "metadata": {},
+   "source": [
+    "## Define the graph\n",
+    "\n",
+    "We can now put it all together and define the graph!"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 154,
+   "id": "812b4e70-4956-4415-8880-db48b3dcbad2",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "from langgraph.graph import StateGraph, END\n",
+    "from langgraph.graph.message import add_messages\n",
+    "from langgraph.managed.few_shot import FewShotExamples\n",
+    "from typing import TypedDict, Annotated\n",
+    "from langchain_core.messages import AnyMessage, HumanMessage, SystemMessage\n",
+    "\n",
+    "class BaseState(TypedDict):\n",
+    "    messages: Annotated[list[AnyMessage], add_messages]\n",
+    "    examples: Annotated[list, FewShotExamples]\n",
+    "\n",
+    "\n",
+    "from langchain_core.messages import AIMessage, ToolMessage\n",
+    "def _render_message(m):\n",
+    "    if isinstance(m, HumanMessage):\n",
+    "        return \"Human: \" + m.content\n",
+    "    elif isinstance(m, AIMessage):\n",
+    "        _m = \"AI: \" + m.content\n",
+    "        if len(m.tool_calls) > 0:\n",
+    "            _m += f\" Tools: {m.tool_calls}\"\n",
+    "        return _m\n",
+    "    elif isinstance(m, ToolMessage):\n",
+    "        return \"Tool Result: ...\"\n",
+    "    else:\n",
+    "        raise ValueError\n",
+    "def _render_messages(ms):\n",
+    "    m_string = [_render_message(m) for m in ms]\n",
+    "    return \"\\n\".join(m_string)\n",
+    "\n",
+    "# Define a new graph\n",
+    "workflow = StateGraph(BaseState)\n",
+    "\n",
+    "def _agent(state: BaseState):\n",
+    "    if len(state['examples']) > 0:\n",
+    "        _examples = \"\\n\\n\".join([f\"Example {i}: \" + _render_messages(e['messages']) for i, e in enumerate(state['examples'])])\n",
+    "        system_message = \"\"\"You are a helpful assistant. Below are some examples of interactions you had with users. \\\n",
+    "These were good interactions where the final result they got was the desired one. As much as possible, you should learn from these interactions and mimic them in the future. \\\n",
+    "Pay particularly close attention to when tools are called, and what the inputs are.!\n",
+    "\n",
+    "{examples}\n",
+    "\n",
+    "Assist the user as they require!\"\"\".format(examples=_examples)\n",
+    "\n",
+    "    else:\n",
+    "        system_message = \"\"\"You are a helpful assistant\"\"\"\n",
+    "    output = model.invoke([SystemMessage(content=system_message)] + state['messages'])\n",
+    "    return {\"messages\": [output]}\n",
+    "\n",
+    "# Define the two nodes we will cycle between\n",
+    "workflow.add_node(\"agent\", _agent)\n",
+    "workflow.add_node(\"action\", tool_node)\n",
+    "\n",
+    "# Set the entrypoint as `agent`\n",
+    "# This means that this node is the first one called\n",
+    "workflow.set_entry_point(\"agent\")\n",
+    "\n",
+    "# We now add a conditional edge\n",
+    "workflow.add_conditional_edges(\n",
+    "    # First, we define the start node. We use `agent`.\n",
+    "    # This means these are the edges taken after the `agent` node is called.\n",
+    "    \"agent\",\n",
+    "    # Next, we pass in the function that will determine which node is called next.\n",
+    "    should_continue,\n",
+    "    # Finally we pass in a mapping.\n",
+    "    # The keys are strings, and the values are other nodes.\n",
+    "    # END is a special node marking that the graph should finish.\n",
+    "    # What will happen is we will call `should_continue`, and then the output of that\n",
+    "    # will be matched against the keys in this mapping.\n",
+    "    # Based on which one it matches, that node will then be called.\n",
+    "    {\n",
+    "        # If `tools`, then we call the tool node.\n",
+    "        \"continue\": \"action\",\n",
+    "        # Otherwise we finish.\n",
+    "        \"end\": END,\n",
+    "    },\n",
+    ")\n",
+    "\n",
+    "# We now add a normal edge from `tools` to `agent`.\n",
+    "# This means that after `tools` is called, `agent` node is called next.\n",
+    "workflow.add_edge(\"action\", \"agent\")"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "bc9c8536-f90b-44fa-958d-5df016c66d8f",
+   "metadata": {},
+   "source": [
+    "**Persistence**\n",
+    "\n",
+    "To add in persistence, we pass in a checkpoint when compiling the graph"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 115,
+   "id": "6845ed6a-d155-4105-9160-28849877248b",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "from langgraph.checkpoint.sqlite import SqliteSaver\n",
+    "\n",
+    "memory = SqliteSaver.from_conn_string(\":memory:\")"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 155,
+   "id": "79d29875-8aa8-434c-9f20-1c58346a6249",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "# Finally, we compile it!\n",
+    "# This compiles it into a LangChain Runnable,\n",
+    "# meaning you can use it as you would any other runnable\n",
+    "app = workflow.compile(checkpointer=memory, interrupt_before=['action'])"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "e8aff75b-563e-42b1-969b-742201514fc3",
+   "metadata": {},
+   "source": [
+    "## Preview the graph"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 156,
+   "id": "c9ab60eb-679b-4eef-9e64-5ffbf3dffc70",
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "image/png": "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",
+      "text/plain": [
+       "<IPython.core.display.Image object>"
+      ]
+     },
+     "execution_count": 156,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "from IPython.display import Image\n",
+    "\n",
+    "Image(app.get_graph().draw_png())"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "2a1b56c5-bd61-4192-8bdb-458a1e9f0159",
+   "metadata": {},
+   "source": [
+    "## Interacting with the Agent\n",
+    "\n",
+    "We can now interact with the agent. Between interactions you can get and update state.\n"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 118,
+   "id": "cfd140f0-a5a6-4697-8115-322242f197b5",
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "{'messages': [AIMessage(content='', additional_kwargs={'tool_calls': [{'id': 'call_yQrJa8CEOfKBdpVl80jzWf5h', 'function': {'arguments': '{\"query\":\"weather in San Francisco\"}', 'name': 'tavily_search_results_json'}, 'type': 'function'}]}, response_metadata={'token_usage': {'completion_tokens': 21, 'prompt_tokens': 94, 'total_tokens': 115}, 'model_name': 'gpt-3.5-turbo', 'system_fingerprint': 'fp_c2295e73ad', 'finish_reason': 'tool_calls', 'logprobs': None}, id='run-3764f79e-17b4-4aa3-bbe4-4f92b11ca52c-0', tool_calls=[{'name': 'tavily_search_results_json', 'args': {'query': 'weather in San Francisco'}, 'id': 'call_yQrJa8CEOfKBdpVl80jzWf5h'}])]}\n"
+     ]
+    }
+   ],
+   "source": [
+    "from langchain_core.messages import HumanMessage\n",
+    "\n",
+    "thread = {\"configurable\": {\"thread_id\": '1'}}\n",
+    "for event in app.stream({\"messages\": [HumanMessage(content=\"whats the weather in sf?\")]}, thread):\n",
+    "    for v in event.values():\n",
+    "        print(v)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 120,
+   "id": "4479f8ae-7c46-4117-8ca2-0a9c2ef9785b",
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "{'messages': [HumanMessage(content='whats the weather in sf?', id='1cfd0c2f-9b60-48da-8938-408fd6aeda13'),\n",
+       "  AIMessage(content='', additional_kwargs={'tool_calls': [{'id': 'call_yQrJa8CEOfKBdpVl80jzWf5h', 'function': {'arguments': '{\"query\":\"weather in San Francisco\"}', 'name': 'tavily_search_results_json'}, 'type': 'function'}]}, response_metadata={'token_usage': {'completion_tokens': 21, 'prompt_tokens': 94, 'total_tokens': 115}, 'model_name': 'gpt-3.5-turbo', 'system_fingerprint': 'fp_c2295e73ad', 'finish_reason': 'tool_calls', 'logprobs': None}, id='run-3764f79e-17b4-4aa3-bbe4-4f92b11ca52c-0', tool_calls=[{'name': 'tavily_search_results_json', 'args': {'query': 'weather in San Francisco'}, 'id': 'call_yQrJa8CEOfKBdpVl80jzWf5h'}])]}"
+      ]
+     },
+     "execution_count": 120,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "current_values = app.get_state(thread)\n",
+    "current_values.values"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 122,
+   "id": "1a0cdb78-40c6-4550-8c27-8f1b02d9e678",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "current_values.values['messages'][-1].tool_calls[0]['args']['query'] = \"weather in San Francisco, Accuweather\""
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 126,
+   "id": "652f699a-89bc-4277-b37a-c3d94b835df5",
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "{'configurable': {'thread_id': '1',\n",
+       "  'thread_ts': '2024-04-20T01:13:15.108790+00:00'}}"
+      ]
+     },
+     "execution_count": 126,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "app.update_state(thread, current_values.values)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 127,
+   "id": "e2b29825-a108-4d40-b377-22e8f4629d64",
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "StateSnapshot(values={'messages': [HumanMessage(content='whats the weather in sf?', id='1cfd0c2f-9b60-48da-8938-408fd6aeda13'), AIMessage(content='', additional_kwargs={'tool_calls': [{'id': 'call_yQrJa8CEOfKBdpVl80jzWf5h', 'function': {'arguments': '{\"query\":\"weather in San Francisco\"}', 'name': 'tavily_search_results_json'}, 'type': 'function'}]}, response_metadata={'token_usage': {'completion_tokens': 21, 'prompt_tokens': 94, 'total_tokens': 115}, 'model_name': 'gpt-3.5-turbo', 'system_fingerprint': 'fp_c2295e73ad', 'finish_reason': 'tool_calls', 'logprobs': None}, id='run-3764f79e-17b4-4aa3-bbe4-4f92b11ca52c-0', tool_calls=[{'name': 'tavily_search_results_json', 'args': {'query': 'weather in San Francisco, Accuweather'}, 'id': 'call_yQrJa8CEOfKBdpVl80jzWf5h'}])]}, next=('action',), config={'configurable': {'thread_id': '1', 'thread_ts': '2024-04-20T01:13:15.108790+00:00'}}, parent_config=None)"
+      ]
+     },
+     "execution_count": 127,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "app.get_state(thread)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 128,
+   "id": "f0aad8a6-056e-42ca-bbc2-b45f768da75f",
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "{'messages': [ToolMessage(content='[{\"url\": \"https://www.weatherapi.com/\", \"content\": \"{\\'location\\': {\\'name\\': \\'San Francisco\\', \\'region\\': \\'California\\', \\'country\\': \\'United States of America\\', \\'lat\\': 37.78, \\'lon\\': -122.42, \\'tz_id\\': \\'America/Los_Angeles\\', \\'localtime_epoch\\': 1713575495, \\'localtime\\': \\'2024-04-19 18:11\\'}, \\'current\\': {\\'last_updated_epoch\\': 1713574800, \\'last_updated\\': \\'2024-04-19 18:00\\', \\'temp_c\\': 16.1, \\'temp_f\\': 61.0, \\'is_day\\': 1, \\'condition\\': {\\'text\\': \\'Sunny\\', \\'icon\\': \\'//cdn.weatherapi.com/weather/64x64/day/113.png\\', \\'code\\': 1000}, \\'wind_mph\\': 16.1, \\'wind_kph\\': 25.9, \\'wind_degree\\': 300, \\'wind_dir\\': \\'WNW\\', \\'pressure_mb\\': 1015.0, \\'pressure_in\\': 29.97, \\'precip_mm\\': 0.0, \\'precip_in\\': 0.0, \\'humidity\\': 67, \\'cloud\\': 0, \\'feelslike_c\\': 16.1, \\'feelslike_f\\': 61.0, \\'vis_km\\': 16.0, \\'vis_miles\\': 9.0, \\'uv\\': 4.0, \\'gust_mph\\': 20.6, \\'gust_kph\\': 33.1}}\"}]', name='tavily_search_results_json', id='8c7e9af3-6569-4982-a83d-be1ec02f828a', tool_call_id='call_yQrJa8CEOfKBdpVl80jzWf5h')]}\n",
+      "{'messages': [AIMessage(content='The current weather in San Francisco is sunny with a temperature of 61.0°F (16.1°C). The wind speed is 25.9 km/h coming from the west-northwest direction. The humidity is at 67%, and there is no precipitation at the moment.', response_metadata={'token_usage': {'completion_tokens': 59, 'prompt_tokens': 476, 'total_tokens': 535}, 'model_name': 'gpt-3.5-turbo', 'system_fingerprint': 'fp_c2295e73ad', 'finish_reason': 'stop', 'logprobs': None}, id='run-7d652a0d-cb00-4857-b9fb-d3e37b0a6d23-0')]}\n"
+     ]
+    }
+   ],
+   "source": [
+    "for event in app.stream(None, thread):\n",
+    "    for v in event.values():\n",
+    "        print(v)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 129,
+   "id": "84748206-975e-4a33-a178-d43df683298c",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "chkpnt_tuple = memory.get_tuple({\"configurable\": {\"thread_id\": \"1\"}})\n",
+    "config = chkpnt_tuple.config\n",
+    "checkpoint = chkpnt_tuple.checkpoint\n",
+    "metadata = chkpnt_tuple.metadata\n",
+    "\n",
+    "# mark as \"good\"\n",
+    "metadata[\"score\"] = 1\n",
+    "memory.put(config, checkpoint, metadata)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 130,
+   "id": "ce7fa228-8c37-4001-afd4-0001b268e1db",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "examples = list(memory.search({\"score\": 1}))"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 131,
+   "id": "0543a501-b4cb-4890-8236-3350ebee5af9",
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "[CheckpointTuple(config={'configurable': {'thread_id': '1', 'thread_ts': '2024-04-20T01:13:36.933600+00:00'}}, checkpoint={'v': 1, 'ts': '2024-04-20T01:13:36.933600+00:00', 'channel_values': {'messages': [HumanMessage(content='whats the weather in sf?', id='1cfd0c2f-9b60-48da-8938-408fd6aeda13'), AIMessage(content='', additional_kwargs={'tool_calls': [{'id': 'call_yQrJa8CEOfKBdpVl80jzWf5h', 'function': {'arguments': '{\"query\":\"weather in San Francisco\"}', 'name': 'tavily_search_results_json'}, 'type': 'function'}]}, response_metadata={'token_usage': {'completion_tokens': 21, 'prompt_tokens': 94, 'total_tokens': 115}, 'model_name': 'gpt-3.5-turbo', 'system_fingerprint': 'fp_c2295e73ad', 'finish_reason': 'tool_calls', 'logprobs': None}, id='run-3764f79e-17b4-4aa3-bbe4-4f92b11ca52c-0', tool_calls=[{'name': 'tavily_search_results_json', 'args': {'query': 'weather in San Francisco, Accuweather'}, 'id': 'call_yQrJa8CEOfKBdpVl80jzWf5h'}]), ToolMessage(content='[{\"url\": \"https://www.weatherapi.com/\", \"content\": \"{\\'location\\': {\\'name\\': \\'San Francisco\\', \\'region\\': \\'California\\', \\'country\\': \\'United States of America\\', \\'lat\\': 37.78, \\'lon\\': -122.42, \\'tz_id\\': \\'America/Los_Angeles\\', \\'localtime_epoch\\': 1713575495, \\'localtime\\': \\'2024-04-19 18:11\\'}, \\'current\\': {\\'last_updated_epoch\\': 1713574800, \\'last_updated\\': \\'2024-04-19 18:00\\', \\'temp_c\\': 16.1, \\'temp_f\\': 61.0, \\'is_day\\': 1, \\'condition\\': {\\'text\\': \\'Sunny\\', \\'icon\\': \\'//cdn.weatherapi.com/weather/64x64/day/113.png\\', \\'code\\': 1000}, \\'wind_mph\\': 16.1, \\'wind_kph\\': 25.9, \\'wind_degree\\': 300, \\'wind_dir\\': \\'WNW\\', \\'pressure_mb\\': 1015.0, \\'pressure_in\\': 29.97, \\'precip_mm\\': 0.0, \\'precip_in\\': 0.0, \\'humidity\\': 67, \\'cloud\\': 0, \\'feelslike_c\\': 16.1, \\'feelslike_f\\': 61.0, \\'vis_km\\': 16.0, \\'vis_miles\\': 9.0, \\'uv\\': 4.0, \\'gust_mph\\': 20.6, \\'gust_kph\\': 33.1}}\"}]', name='tavily_search_results_json', id='8c7e9af3-6569-4982-a83d-be1ec02f828a', tool_call_id='call_yQrJa8CEOfKBdpVl80jzWf5h'), AIMessage(content='The current weather in San Francisco is sunny with a temperature of 61.0°F (16.1°C). The wind speed is 25.9 km/h coming from the west-northwest direction. The humidity is at 67%, and there is no precipitation at the moment.', response_metadata={'token_usage': {'completion_tokens': 59, 'prompt_tokens': 476, 'total_tokens': 535}, 'model_name': 'gpt-3.5-turbo', 'system_fingerprint': 'fp_c2295e73ad', 'finish_reason': 'stop', 'logprobs': None}, id='run-7d652a0d-cb00-4857-b9fb-d3e37b0a6d23-0')], 'agent': {'messages': [AIMessage(content='The current weather in San Francisco is sunny with a temperature of 61.0°F (16.1°C). The wind speed is 25.9 km/h coming from the west-northwest direction. The humidity is at 67%, and there is no precipitation at the moment.', response_metadata={'token_usage': {'completion_tokens': 59, 'prompt_tokens': 476, 'total_tokens': 535}, 'model_name': 'gpt-3.5-turbo', 'system_fingerprint': 'fp_c2295e73ad', 'finish_reason': 'stop', 'logprobs': None}, id='run-7d652a0d-cb00-4857-b9fb-d3e37b0a6d23-0')]}}, 'channel_versions': defaultdict(<class 'int'>, {'__start__': 1, 'messages': 6, 'start:agent': 2, 'action': 5, 'agent': 6, 'branch:agent:should_continue:action': 4}), 'versions_seen': defaultdict(<function _seen_dict at 0x10832bce0>, {'__start__': defaultdict(<class 'int'>, {'__start__': 1}), 'agent': defaultdict(<class 'int'>, {'start:agent': 2, 'action': 5}), 'action': defaultdict(<class 'int'>, {'branch:agent:should_continue:action': 4}), '__interrupt__': defaultdict(<class 'int'>, {'messages': 4})})}, parent_config={'configurable': {'thread_id': '1', 'thread_ts': '2024-04-20T01:13:35.392072+00:00'}})]"
+      ]
+     },
+     "execution_count": 131,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "examples"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 157,
+   "id": "336a70d3-d8c7-4310-a373-df2be3320030",
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "{'messages': [AIMessage(content='', additional_kwargs={'tool_calls': [{'id': 'call_amNrYLgitup6hCDiHUYKwodH', 'function': {'arguments': '{\"query\":\"weather in Los Angeles, Accuweather\"}', 'name': 'tavily_search_results_json'}, 'type': 'function'}]}, response_metadata={'token_usage': {'completion_tokens': 25, 'prompt_tokens': 296, 'total_tokens': 321}, 'model_name': 'gpt-3.5-turbo', 'system_fingerprint': 'fp_c2295e73ad', 'finish_reason': 'tool_calls', 'logprobs': None}, id='run-6b852685-e84f-48e7-b8d0-5b0a44ac9776-0', tool_calls=[{'name': 'tavily_search_results_json', 'args': {'query': 'weather in Los Angeles, Accuweather'}, 'id': 'call_amNrYLgitup6hCDiHUYKwodH'}])]}\n"
+     ]
+    }
+   ],
+   "source": [
+    "thread = {\"configurable\": {\"thread_id\": '7'}}\n",
+    "for event in app.stream({\"messages\": [HumanMessage(content=\"whats the weather in la?\")]}, thread):\n",
+    "    for v in event.values():\n",
+    "        print(v)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 158,
+   "id": "ab245e7c-e47d-45e4-8b79-361fab359e76",
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "{'messages': [ToolMessage(content='[{\"url\": \"https://www.weatherapi.com/\", \"content\": \"{\\'location\\': {\\'name\\': \\'Los Angeles\\', \\'region\\': \\'California\\', \\'country\\': \\'United States of America\\', \\'lat\\': 34.05, \\'lon\\': -118.24, \\'tz_id\\': \\'America/Los_Angeles\\', \\'localtime_epoch\\': 1713576177, \\'localtime\\': \\'2024-04-19 18:22\\'}, \\'current\\': {\\'last_updated_epoch\\': 1713575700, \\'last_updated\\': \\'2024-04-19 18:15\\', \\'temp_c\\': 17.8, \\'temp_f\\': 64.0, \\'is_day\\': 1, \\'condition\\': {\\'text\\': \\'Partly cloudy\\', \\'icon\\': \\'//cdn.weatherapi.com/weather/64x64/day/116.png\\', \\'code\\': 1003}, \\'wind_mph\\': 4.3, \\'wind_kph\\': 6.8, \\'wind_degree\\': 250, \\'wind_dir\\': \\'WSW\\', \\'pressure_mb\\': 1014.0, \\'pressure_in\\': 29.94, \\'precip_mm\\': 0.0, \\'precip_in\\': 0.0, \\'humidity\\': 65, \\'cloud\\': 50, \\'feelslike_c\\': 17.8, \\'feelslike_f\\': 64.0, \\'vis_km\\': 16.0, \\'vis_miles\\': 9.0, \\'uv\\': 5.0, \\'gust_mph\\': 10.3, \\'gust_kph\\': 16.6}}\"}]', name='tavily_search_results_json', id='6477bd73-bdf1-46c5-ab0a-cc808b1a183e', tool_call_id='call_amNrYLgitup6hCDiHUYKwodH')]}\n",
+      "{'messages': [AIMessage(content='The current weather in Los Angeles is partly cloudy with a temperature of 64.0°F (17.8°C). The wind speed is 6.8 km/h coming from the west-southwest direction. The humidity is at 65%, and there is no precipitation at the moment.', response_metadata={'token_usage': {'completion_tokens': 60, 'prompt_tokens': 679, 'total_tokens': 739}, 'model_name': 'gpt-3.5-turbo', 'system_fingerprint': 'fp_c2295e73ad', 'finish_reason': 'stop', 'logprobs': None}, id='run-a604c04e-c3c3-4545-804f-a89aee6bf516-0')]}\n"
+     ]
+    }
+   ],
+   "source": [
+    "for event in app.stream(None, thread):\n",
+    "    for v in event.values():\n",
+    "        print(v)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "9ab115de-9b11-4e8b-8ace-c23e1369300b",
+   "metadata": {},
+   "outputs": [],
+   "source": []
+  }
+ ],
+ "metadata": {
+  "kernelspec": {
+   "display_name": "Python 3 (ipykernel)",
+   "language": "python",
+   "name": "python3"
+  },
+  "language_info": {
+   "codemirror_mode": {
+    "name": "ipython",
+    "version": 3
+   },
+   "file_extension": ".py",
+   "mimetype": "text/x-python",
+   "name": "python",
+   "nbconvert_exporter": "python",
+   "pygments_lexer": "ipython3",
+   "version": "3.11.9"
+  }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 5
+}
diff --git a/langgraph/checkpoint/aiosqlite.py b/langgraph/checkpoint/aiosqlite.py
index 76676e64e..b82cdfad8 100644
--- a/langgraph/checkpoint/aiosqlite.py
+++ b/langgraph/checkpoint/aiosqlite.py
@@ -14,7 +14,7 @@ from langgraph.checkpoint.base import (
     CheckpointTuple,
     SerializerProtocol,
 )
-from langgraph.checkpoint.sqlite import JsonPlusSerializerCompat
+from langgraph.checkpoint.sqlite import JsonPlusSerializerCompat, search_where
 
 
 class AsyncSqliteSaver(BaseCheckpointSaver, AbstractAsyncContextManager):
@@ -252,6 +252,50 @@ class AsyncSqliteSaver(BaseCheckpointSaver, AbstractAsyncContextManager):
                     else None,
                 )
 
+    async def asearch(
+        self,
+        metadata_filter: CheckpointMetadata,
+        *,
+        before: Optional[RunnableConfig] = None,
+        limit: Optional[int] = None,
+    ) -> AsyncIterator[CheckpointTuple]:
+        """Search for checkpoints by metadata asynchronously.
+
+        This method retrieves a list of checkpoint tuples from the SQLite
+        database based on the provided metadata filter. The metadata filter does
+        not need to contain all keys defined in the CheckpointMetadata class.
+        The checkpoints are ordered by timestamp in descending order.
+
+        Args:
+            metadata_filter (CheckpointMetadata): The metadata filter to use for searching the checkpoints.
+            before (Optional[RunnableConfig]): If provided, only checkpoints before the specified timestamp are returned. Defaults to None.
+            limit (Optional[int]): The maximum number of checkpoints to return. Defaults to None.
+
+        Yields:
+            Iterator[CheckpointTuple]: An iterator of checkpoint tuples.
+        """
+        await self.setup()
+
+        # construct query
+        SELECT = "SELECT thread_id, thread_ts, parent_ts, checkpoint, metadata FROM checkpoints "
+        WHERE, params = search_where(metadata_filter, before)
+        ORDER_BY = "ORDER BY thread_ts DESC "
+        LIMIT = f"LIMIT {limit}" if limit else ""
+
+        query = f"{SELECT}{WHERE}{ORDER_BY}{LIMIT}"
+
+        # execute query
+        async with self.conn.execute(query, params) as cursor:
+            async for thread_id, thread_ts, parent_ts, value, metadata in cursor:
+                yield CheckpointTuple(
+                    {"configurable": {"thread_id": thread_id, "thread_ts": thread_ts}},
+                    self.serde.loads(value),
+                    self.serde.loads(metadata) if metadata is not None else {},
+                    {"configurable": {"thread_id": thread_id, "thread_ts": parent_ts}}
+                    if parent_ts
+                    else None,
+                )
+
     async def aput(
         self,
         config: RunnableConfig,
diff --git a/langgraph/checkpoint/base.py b/langgraph/checkpoint/base.py
index 695eb3d96..5a9dc99ce 100644
--- a/langgraph/checkpoint/base.py
+++ b/langgraph/checkpoint/base.py
@@ -36,6 +36,11 @@ class CheckpointMetadata(TypedDict, total=False):
 
     Mapping from node name to writes emitted by that node.
     """
+    score: Optional[int]
+    """The score of the checkpoint.
+    
+    The score can be used to mark a checkpoint as "good".
+    """
 
 
 class Checkpoint(TypedDict):
@@ -148,6 +153,15 @@ class BaseCheckpointSaver(ABC):
     ) -> Iterator[CheckpointTuple]:
         raise NotImplementedError
 
+    def search(
+        self,
+        metadata_filter: CheckpointMetadata,
+        *,
+        before: Optional[RunnableConfig] = None,
+        limit: Optional[int] = None,
+    ) -> Iterator[CheckpointTuple]:
+        raise NotImplementedError
+
     def put(
         self,
         config: RunnableConfig,
@@ -173,6 +187,16 @@ class BaseCheckpointSaver(ABC):
         raise NotImplementedError
         yield
 
+    def asearch(
+        self,
+        metadata_filter: CheckpointMetadata,
+        *,
+        before: Optional[RunnableConfig] = None,
+        limit: Optional[int] = None,
+    ) -> AsyncIterator[CheckpointTuple]:
+        raise NotImplementedError
+        yield
+
     async def aput(
         self,
         config: RunnableConfig,
diff --git a/langgraph/checkpoint/memory.py b/langgraph/checkpoint/memory.py
index ec8316e99..f8c49ce78 100644
--- a/langgraph/checkpoint/memory.py
+++ b/langgraph/checkpoint/memory.py
@@ -121,6 +121,57 @@ class MemorySaver(BaseCheckpointSaver):
                 metadata=self.serde.loads(metadata),
             )
 
+    def search(
+        self,
+        metadata_filter: CheckpointMetadata,
+        *,
+        before: Optional[RunnableConfig] = None,
+        limit: Optional[int] = None,
+    ) -> Iterator[CheckpointTuple]:
+        """Search for checkpoints by metadata.
+
+        This method retrieves a list of checkpoint tuples from the in-memory
+        storage based on the provided metadata filter. The metadata filter does
+        not need to contain all keys defined in the CheckpointMetadata class.
+        The checkpoints are ordered by timestamp in descending order.
+
+        Args:
+            metadata_filter (CheckpointMetadata): The metadata filter to use for searching the checkpoints.
+            before (Optional[RunnableConfig]): If provided, only checkpoints before the specified timestamp are returned. Defaults to None.
+            limit (Optional[int]): The maximum number of checkpoints to return. Defaults to None.
+
+        Yields:
+            Iterator[CheckpointTuple]: An iterator of checkpoint tuples.
+        """
+        for thread_id, checkpoints in self.storage.items():
+            for ts, (checkpoint_bytes, metadata_bytes) in checkpoints.items():
+                # filter by thread_ts
+                if before and ts >= before["configurable"]["thread_ts"]:
+                    continue
+
+                # check if all query key/value pairs match the metadata
+                metadata = self.serde.loads(metadata_bytes)
+                all_keys_match = all(
+                    query_value == metadata[query_key]
+                    for query_key, query_value in metadata_filter.items()
+                )
+
+                # if all query key/value pairs match, yield the checkpoint
+                if all_keys_match:
+                    # limit search results
+                    if limit is not None:
+                        if limit <= 0:
+                            break
+                        limit -= 1
+
+                    yield CheckpointTuple(
+                        config={
+                            "configurable": {"thread_id": thread_id, "thread_ts": ts}
+                        },
+                        checkpoint=self.serde.loads(checkpoint_bytes),
+                        metadata=metadata,
+                    )
+
     def put(
         self,
         config: RunnableConfig,
@@ -200,6 +251,29 @@ class MemorySaver(BaseCheckpointSaver):
             else:
                 break
 
+    async def asearch(
+        self,
+        metadata_filter: CheckpointMetadata,
+        *,
+        before: Optional[RunnableConfig] = None,
+        limit: Optional[int] = None,
+    ) -> AsyncIterator[CheckpointTuple]:
+        """Asynchronous version of search.
+
+        This method is an asynchronous wrapper around search that runs the synchronous
+        method in a separate thread using asyncio.
+        """
+        loop = asyncio.get_running_loop()
+        iter = await loop.run_in_executor(
+            None, partial(self.search, before=before, limit=limit), metadata_filter
+        )
+
+        while True:
+            if item := await loop.run_in_executor(None, next, iter, None):
+                yield item
+            else:
+                break
+
     async def aput(
         self,
         config: RunnableConfig,
diff --git a/langgraph/checkpoint/sqlite.py b/langgraph/checkpoint/sqlite.py
index a34e295c0..e5835f17b 100644
--- a/langgraph/checkpoint/sqlite.py
+++ b/langgraph/checkpoint/sqlite.py
@@ -1,9 +1,10 @@
+import json
 import pickle
 import sqlite3
 import threading
 from contextlib import AbstractContextManager, contextmanager
 from types import TracebackType
-from typing import Any, Iterator, Optional
+from typing import Any, Iterator, Optional, Tuple
 
 from langchain_core.runnables import RunnableConfig
 from typing_extensions import Self
@@ -339,6 +340,57 @@ class SqliteSaver(BaseCheckpointSaver, AbstractContextManager):
                     ),
                 )
 
+    def search(
+        self,
+        metadata_filter: CheckpointMetadata,
+        *,
+        before: Optional[RunnableConfig] = None,
+        limit: Optional[int] = None,
+    ) -> Iterator[CheckpointTuple]:
+        """Search for checkpoints by metadata.
+
+        This method retrieves a list of checkpoint tuples from the SQLite
+        database based on the provided metadata filter. The metadata filter does
+        not need to contain all keys defined in the CheckpointMetadata class.
+        The checkpoints are ordered by timestamp in descending order.
+
+        Args:
+            metadata_filter (CheckpointMetadata): The metadata filter to use for searching the checkpoints.
+            before (Optional[RunnableConfig]): If provided, only checkpoints before the specified timestamp are returned. Defaults to None.
+            limit (Optional[int]): The maximum number of checkpoints to return. Defaults to None.
+
+        Yields:
+            Iterator[CheckpointTuple]: An iterator of checkpoint tuples.
+        """
+        # construct query
+        SELECT = "SELECT thread_id, thread_ts, parent_ts, checkpoint, metadata FROM checkpoints "
+        WHERE, params = search_where(metadata_filter, before)
+        ORDER_BY = "ORDER BY thread_ts DESC "
+        LIMIT = f"LIMIT {limit}" if limit else ""
+
+        query = f"{SELECT}{WHERE}{ORDER_BY}{LIMIT}"
+
+        # execute query
+        with self.cursor(transaction=False) as cur:
+            cur.execute(query, params)
+
+            for thread_id, thread_ts, parent_ts, value, metadata in cur:
+                yield CheckpointTuple(
+                    {"configurable": {"thread_id": thread_id, "thread_ts": thread_ts}},
+                    self.serde.loads(value),
+                    self.serde.loads(metadata) if metadata is not None else {},
+                    (
+                        {
+                            "configurable": {
+                                "thread_id": thread_id,
+                                "thread_ts": parent_ts,
+                            }
+                        }
+                        if parent_ts
+                        else None
+                    ),
+                )
+
     def put(
         self,
         config: RunnableConfig,
@@ -386,3 +438,89 @@ class SqliteSaver(BaseCheckpointSaver, AbstractContextManager):
                 "thread_ts": checkpoint["ts"],
             }
         }
+
+
+def search_where(
+    metadata_filter: CheckpointMetadata,
+    before: Optional[RunnableConfig] = None,
+) -> Tuple[str, Tuple[Any, ...]]:
+    """Return WHERE clause predicates for (a)search() given metadata filter
+    and `before` config.
+
+    This method returns a tuple of a string and a tuple of values. The string
+    is the parametered WHERE clause predicate (including the WHERE keyword):
+    "WHERE column1 = ? AND column2 IS ?". The tuple of values contains the
+    values for each of the corresponding parameters.
+    """
+    where = "WHERE "
+    param_values = ()
+
+    # construct predicate for metadata filter
+    metadata_predicate, metadata_values = _metadata_predicate(metadata_filter)
+    if metadata_predicate != "":
+        where += metadata_predicate
+        param_values += metadata_values
+
+    # construct predicate for `before`
+    if before is not None:
+        if metadata_predicate != "":
+            where += "AND thread_ts < ? "
+        else:
+            where += "thread_ts < ? "
+
+        param_values += (before["configurable"]["thread_ts"],)
+
+    if where == "WHERE ":
+        # no predicates, return an empty WHERE clause string
+        return ("", ())
+    else:
+        return (where, param_values)
+
+
+def _metadata_predicate(
+    metadata_filter: CheckpointMetadata,
+) -> Tuple[str, Tuple[Any, ...]]:
+    """Return WHERE clause predicates for (a)search() given metadata filter.
+
+    This method returns a tuple of a string and a tuple of values. The string
+    is the parametered WHERE clause predicate (excluding the WHERE keyword):
+    "column1 = ? AND column2 IS ?". The tuple of values contains the values
+    for each of the corresponding parameters.
+    """
+
+    def _where_value(query_value: Any) -> Tuple[str, Any]:
+        """Return tuple of operator and value for WHERE clause predicate."""
+        if query_value is None:
+            return ("IS ?", None)
+        elif (
+            isinstance(query_value, str)
+            or isinstance(query_value, int)
+            or isinstance(query_value, float)
+        ):
+            return ("= ?", query_value)
+        elif isinstance(query_value, bool):
+            return ("= ?", 1 if query_value else 0)
+        elif isinstance(query_value, dict) or isinstance(query_value, list):
+            # query value for JSON object cannot have trailing space after separators (, :)
+            # SQLite json_extract() returns JSON string without whitespace
+            return ("= ?", json.dumps(query_value, separators=(",", ":")))
+        else:
+            return ("= ?", str(query_value))
+
+    predicate = ""
+    param_values = ()
+
+    # process metadata query
+    for query_key, query_value in metadata_filter.items():
+        operator, param_value = _where_value(query_value)
+        predicate += (
+            f"json_extract(CAST(metadata AS TEXT), '$.{query_key}') {operator} AND "
+        )
+        param_values += (param_value,)
+
+    if predicate != "":
+        # remove trailing AND
+        predicate = predicate[:-4]
+
+    # predicate contains an extra trailing space
+    return (predicate, param_values)
diff --git a/langgraph/managed/base.py b/langgraph/managed/base.py
index 15b47070b..b06f6a7c7 100644
--- a/langgraph/managed/base.py
+++ b/langgraph/managed/base.py
@@ -8,9 +8,10 @@ from typing import (
     AsyncGenerator,
     Generator,
     Generic,
-    Sequence,
+    NamedTuple,
     Type,
     TypeVar,
+    Union,
 )
 
 from langchain_core.runnables import RunnableConfig
@@ -32,10 +33,10 @@ class ManagedValue(ABC, Generic[V]):
     @classmethod
     @contextmanager
     def enter(
-        cls, config: RunnableConfig, graph: "Pregel"
+        cls, config: RunnableConfig, graph: "Pregel", **kwargs: Any
     ) -> Generator[Self, None, None]:
         try:
-            value = cls(config, graph)
+            value = cls(config, graph, **kwargs)
             yield value
         finally:
             # because managed value and Pregel have reference to each other
@@ -48,10 +49,10 @@ class ManagedValue(ABC, Generic[V]):
     @classmethod
     @asynccontextmanager
     async def aenter(
-        cls, config: RunnableConfig, graph: "Pregel"
+        cls, config: RunnableConfig, graph: "Pregel", **kwargs: Any
     ) -> AsyncGenerator[Self, None]:
         try:
-            value = cls(config, graph)
+            value = cls(config, graph, **kwargs)
             yield value
         finally:
             # because managed value and Pregel have reference to each other
@@ -66,40 +67,64 @@ class ManagedValue(ABC, Generic[V]):
         ...
 
 
-def is_managed_value(value: Any) -> TypeGuard[Type[ManagedValue]]:
-    return isclass(value) and issubclass(value, ManagedValue)
+class ConfiguredManagedValue(NamedTuple):
+    cls: Type[ManagedValue]
+    kwargs: dict[str, Any]
+
+
+ManagedValueSpec = Union[Type[ManagedValue], ConfiguredManagedValue]
+
+ManagedValueMapping = dict[str, ManagedValue]
+
+
+def is_managed_value(value: Any) -> TypeGuard[ManagedValueSpec]:
+    return (isclass(value) and issubclass(value, ManagedValue)) or isinstance(
+        value, ConfiguredManagedValue
+    )
 
 
 @contextmanager
 def ManagedValuesManager(
-    values: Sequence[Type[ManagedValue]],
+    values: dict[str, ManagedValueSpec],
     config: RunnableConfig,
     graph: "Pregel",
-) -> Generator[Sequence[ManagedValue], None, None]:
-    with ExitStack() as stack:
-        unique: list[Type[ManagedValue]] = []
-        for value in values:
-            if value not in unique:
-                unique.append(value)
-
-        yield [stack.enter_context(value.enter(config, graph)) for value in unique]
+) -> Generator[ManagedValueMapping, None, None]:
+    if values:
+        with ExitStack() as stack:
+            yield {
+                key: stack.enter_context(
+                    value.cls.enter(config, graph, **value.kwargs)
+                    if isinstance(value, ConfiguredManagedValue)
+                    else value.enter(config, graph)
+                )
+                for key, value in values.items()
+            }
+    else:
+        yield {}
 
 
 @asynccontextmanager
 async def AsyncManagedValuesManager(
-    values: Sequence[Type[ManagedValue]],
+    values: dict[str, ManagedValueSpec],
     config: RunnableConfig,
     graph: "Pregel",
-) -> AsyncGenerator[Sequence[ManagedValue], None]:
-    async with AsyncExitStack() as stack:
-        unique: list[Type[ManagedValue]] = []
-        for value in values:
-            if value not in unique:
-                unique.append(value)
-
-        yield await asyncio.gather(
-            *(
-                stack.enter_async_context(value.aenter(config, graph))
-                for value in unique
-            )
-        )
+) -> AsyncGenerator[ManagedValueMapping, None]:
+    if values:
+        async with AsyncExitStack() as stack:
+            # create enter tasks with reference to spec
+            tasks = {
+                asyncio.create_task(
+                    stack.enter_async_context(
+                        value.cls.aenter(config, graph, **value.kwargs)
+                        if isinstance(value, ConfiguredManagedValue)
+                        else value.aenter(config, graph)
+                    )
+                ): key
+                for key, value in values.items()
+            }
+            # wait for all enter tasks
+            done, _ = await asyncio.wait(tasks, return_when=asyncio.ALL_COMPLETED)
+            # build mapping from spec to result
+            yield {tasks[task]: task.result() for task in done}
+    else:
+        yield {}
diff --git a/langgraph/managed/few_shot.py b/langgraph/managed/few_shot.py
new file mode 100644
index 000000000..d7cf399f4
--- /dev/null
+++ b/langgraph/managed/few_shot.py
@@ -0,0 +1,87 @@
+from contextlib import asynccontextmanager, contextmanager
+from typing import (
+    TYPE_CHECKING,
+    Any,
+    AsyncGenerator,
+    AsyncIterator,
+    Generator,
+    Generic,
+    Iterator,
+    Sequence,
+)
+
+from langchain_core.runnables import RunnableConfig
+from typing_extensions import Self
+
+from langgraph.channels.base import AsyncChannelsManager, ChannelsManager
+from langgraph.managed.base import ConfiguredManagedValue, ManagedValue, V
+from langgraph.pregel import Pregel
+from langgraph.pregel.io import read_channels
+from langgraph.pregel.types import PregelTaskDescription
+
+if TYPE_CHECKING:
+    from langgraph.pregel import Pregel
+
+
+class FewShotExamples(ManagedValue[Sequence[V]], Generic[V]):
+    examples: list[V]
+
+    def __init__(
+        self,
+        config: RunnableConfig,
+        graph: Pregel,
+        k: int = 5,
+        metadata_filter: dict[str, Any] = None,
+    ) -> None:
+        super().__init__(config, graph)
+        self.k = k
+        self.metadata_filter = metadata_filter or {}
+
+    @classmethod
+    def configure(
+        cls, k: int = 5, metadata_filter: dict[str, Any] = None
+    ) -> ConfiguredManagedValue:
+        return ConfiguredManagedValue(
+            cls,
+            {
+                "k": k,
+                "metadata_filter": metadata_filter,
+            },
+        )
+
+    def iter(self, score: int = 1) -> Iterator[V]:
+        for example in self.graph.checkpointer.search(
+            {"score": score, **self.metadata_filter}, limit=self.k
+        ):
+            with ChannelsManager(self.graph.channels, example.checkpoint) as channels:
+                yield read_channels(channels, self.graph.output_channels)
+
+    async def aiter(self, score: int = 1) -> AsyncIterator[V]:
+        async for example in self.graph.checkpointer.asearch(
+            {"score": score, **self.metadata_filter}, limit=self.k
+        ):
+            async with AsyncChannelsManager(
+                self.graph.channels, example.checkpoint
+            ) as channels:
+                yield read_channels(channels, self.graph.output_channels)
+
+    @classmethod
+    @contextmanager
+    def enter(
+        cls, config: RunnableConfig, graph: "Pregel", **kwargs: Any
+    ) -> Generator[Self, None, None]:
+        with super().enter(config, graph, **kwargs) as value:
+            value.examples = list(value.iter())
+            yield value
+
+    @classmethod
+    @asynccontextmanager
+    async def aenter(
+        cls, config: RunnableConfig, graph: "Pregel", **kwargs: Any
+    ) -> AsyncGenerator[Self, None]:
+        async with super().aenter(config, graph, **kwargs) as value:
+            value.examples = [e async for e in value.aiter()]
+            yield value
+
+    def __call__(self, step: int, task: PregelTaskDescription) -> Sequence[V]:
+        return self.examples
diff --git a/langgraph/pregel/__init__.py b/langgraph/pregel/__init__.py
index ec7687fd8..63a78833f 100644
--- a/langgraph/pregel/__init__.py
+++ b/langgraph/pregel/__init__.py
@@ -4,7 +4,6 @@ import asyncio
 import concurrent.futures
 from collections import defaultdict, deque
 from functools import partial
-from inspect import isclass
 from typing import (
     Any,
     AsyncIterator,
@@ -70,8 +69,9 @@ from langgraph.constants import (
 from langgraph.errors import GraphRecursionError, InvalidUpdateError
 from langgraph.managed.base import (
     AsyncManagedValuesManager,
-    ManagedValue,
+    ManagedValueMapping,
     ManagedValuesManager,
+    ManagedValueSpec,
     is_managed_value,
 )
 from langgraph.pregel.debug import (
@@ -324,14 +324,14 @@ class Pregel(
         return self.stream_channels or [k for k in self.channels]
 
     @property
-    def managed_values_list(self) -> Sequence[Type[ManagedValue]]:
-        return [
-            v
+    def managed_values_dict(self) -> dict[str, ManagedValueSpec]:
+        return {
+            k: v
             for node in self.nodes.values()
             if isinstance(node.channels, dict)
-            for v in node.channels.values()
+            for k, v in node.channels.items()
             if is_managed_value(v)
-        ]
+        }
 
     def get_state(self, config: RunnableConfig) -> StateSnapshot:
         """Get the current state of the graph."""
@@ -344,7 +344,7 @@ class Pregel(
         with ChannelsManager(
             self.channels, checkpoint
         ) as channels, ManagedValuesManager(
-            self.managed_values_list, ensure_config(config), self
+            self.managed_values_dict, ensure_config(config), self
         ) as managed:
             _, next_tasks = _prepare_next_tasks(
                 checkpoint,
@@ -375,7 +375,7 @@ class Pregel(
         async with AsyncChannelsManager(
             self.channels, checkpoint
         ) as channels, AsyncManagedValuesManager(
-            self.managed_values_list, ensure_config(config), self
+            self.managed_values_dict, ensure_config(config), self
         ) as managed:
             _, next_tasks = _prepare_next_tasks(
                 checkpoint,
@@ -411,7 +411,7 @@ class Pregel(
             with ChannelsManager(
                 self.channels, checkpoint
             ) as channels, ManagedValuesManager(
-                self.managed_values_list, ensure_config(config), self
+                self.managed_values_dict, ensure_config(config), self
             ) as managed:
                 _, next_tasks = _prepare_next_tasks(
                     checkpoint,
@@ -450,7 +450,7 @@ class Pregel(
             async with AsyncChannelsManager(
                 self.channels, checkpoint
             ) as channels, AsyncManagedValuesManager(
-                self.managed_values_list, ensure_config(config), self
+                self.managed_values_dict, ensure_config(config), self
             ) as managed:
                 _, next_tasks = _prepare_next_tasks(
                     checkpoint,
@@ -722,7 +722,7 @@ class Pregel(
             ) as channels, get_executor_for_config(
                 config
             ) as executor, ManagedValuesManager(
-                self.managed_values_list, config, self
+                self.managed_values_dict, config, self
             ) as managed:
                 # map inputs to channel updates
                 if input_writes := deque(map_input(input_keys, input)):
@@ -1018,7 +1018,7 @@ class Pregel(
             async with AsyncChannelsManager(
                 self.channels, checkpoint
             ) as channels, AsyncManagedValuesManager(
-                self.managed_values_list, config, self
+                self.managed_values_dict, config, self
             ) as managed:
                 # map inputs to channel updates
                 if input_writes := deque(map_input(input_keys, input)):
@@ -1474,7 +1474,7 @@ def _prepare_next_tasks(
     checkpoint: Checkpoint,
     processes: Mapping[str, PregelNode],
     channels: Mapping[str, BaseChannel],
-    managed: Sequence[ManagedValue],
+    managed: ManagedValueMapping,
     config: RunnableConfig,
     step: int,
     for_execution: Literal[False],
@@ -1487,7 +1487,7 @@ def _prepare_next_tasks(
     checkpoint: Checkpoint,
     processes: Mapping[str, PregelNode],
     channels: Mapping[str, BaseChannel],
-    managed: Sequence[ManagedValue],
+    managed: ManagedValueMapping,
     config: RunnableConfig,
     step: int,
     for_execution: Literal[True],
@@ -1499,7 +1499,7 @@ def _prepare_next_tasks(
     checkpoint: Checkpoint,
     processes: Mapping[str, PregelNode],
     channels: Mapping[str, BaseChannel],
-    managed: Sequence[ManagedValue],
+    managed: ManagedValueMapping,
     config: RunnableConfig,
     step: int,
     *,
@@ -1532,11 +1532,10 @@ def _prepare_next_tasks(
 
                     managed_values = {}
                     for key, chan in proc.channels.items():
-                        for mv in managed:
-                            if isclass(chan) and isinstance(mv, chan):
-                                managed_values[key] = mv(
-                                    step, PregelTaskDescription(name, val)
-                                )
+                        if is_managed_value(chan):
+                            managed_values[key] = managed[key](
+                                step, PregelTaskDescription(name, val)
+                            )
 
                     val.update(managed_values)
                 except EmptyChannelError:
diff --git a/langgraph/pregel/read.py b/langgraph/pregel/read.py
index e8576e22f..eefed2b47 100644
--- a/langgraph/pregel/read.py
+++ b/langgraph/pregel/read.py
@@ -1,6 +1,6 @@
 from __future__ import annotations
 
-from typing import Any, Callable, Mapping, Optional, Sequence, Type, Union
+from typing import Any, Callable, Mapping, Optional, Sequence, Union
 
 from langchain_core.pydantic_v1 import Field
 from langchain_core.runnables import (
@@ -15,7 +15,7 @@ from langchain_core.runnables.config import merge_configs
 from langchain_core.runnables.utils import ConfigurableFieldSpec
 
 from langgraph.constants import CONFIG_KEY_READ
-from langgraph.managed.base import ManagedValue
+from langgraph.managed.base import ManagedValueSpec
 from langgraph.pregel.write import ChannelWrite
 from langgraph.utils import RunnableCallable
 
@@ -100,7 +100,7 @@ DEFAULT_BOUND: RunnablePassthrough = RunnablePassthrough()
 
 
 class PregelNode(RunnableBindingBase):
-    channels: Union[list[str], Mapping[str, Union[str, Type[ManagedValue]]]]
+    channels: Union[list[str], Mapping[str, Union[str, ManagedValueSpec]]]
 
     triggers: list[str] = Field(default_factory=list)
 
diff --git a/tests/checkpoint/__init__.py b/tests/checkpoint/__init__.py
new file mode 100644
index 000000000..e69de29bb
diff --git a/tests/checkpoint/test_aiosqlite.py b/tests/checkpoint/test_aiosqlite.py
new file mode 100644
index 000000000..ef06efdcd
--- /dev/null
+++ b/tests/checkpoint/test_aiosqlite.py
@@ -0,0 +1,79 @@
+import pytest
+from langchain_core.runnables import RunnableConfig
+
+from langgraph.checkpoint.aiosqlite import AsyncSqliteSaver
+from langgraph.checkpoint.base import Checkpoint, CheckpointMetadata
+
+
+class TestAsyncSqliteSaver:
+    @pytest.fixture(autouse=True)
+    def setup(self):
+        self.sqlite_saver = AsyncSqliteSaver.from_conn_string(":memory:")
+
+        # objects for test setup
+        self.config_1: RunnableConfig = {
+            "configurable": {"thread_id": "thread-1", "thread_ts": "1"}
+        }
+        self.config_2: RunnableConfig = {
+            "configurable": {"thread_id": "thread-2", "thread_ts": "2"}
+        }
+
+        self.chkpnt_1: Checkpoint = {
+            "v": 1,
+            "ts": "1",
+            "channel_values": {},
+            "channel_versions": {},
+            "versions_seen": {},
+        }
+        self.chkpnt_2: Checkpoint = {
+            "v": 2,
+            "ts": "2",
+            "channel_values": {},
+            "channel_versions": {},
+            "versions_seen": {},
+        }
+
+        self.metadata_1: CheckpointMetadata = {
+            "source": "input",
+            "step": 2,
+            "writes": {},
+            "score": 1,
+        }
+        self.metadata_2: CheckpointMetadata = {
+            "source": "loop",
+            "step": 1,
+            "writes": {"foo": "bar"},
+            "score": None,
+        }
+
+    async def test_asearch(self):
+        # set up test
+        # save checkpoints
+        await self.sqlite_saver.aput(self.config_1, self.chkpnt_1, self.metadata_1)
+        await self.sqlite_saver.aput(self.config_2, self.chkpnt_2, self.metadata_2)
+
+        # call method / assertions
+        query_1: CheckpointMetadata = {"source": "input"}  # search by 1 key
+        query_2: CheckpointMetadata = {
+            "step": 1,
+            "writes": {"foo": "bar"},
+        }  # search by multiple keys
+        query_3: CheckpointMetadata = {}  # search by no keys, return all checkpoints
+        query_4: CheckpointMetadata = {"source": "update", "step": 1}  # no match
+
+        async with self.sqlite_saver as sqlite_saver:
+            search_results_1 = [c async for c in sqlite_saver.asearch(query_1)]
+            assert len(search_results_1) == 1
+            assert search_results_1[0].metadata == self.metadata_1
+
+            search_results_2 = [c async for c in sqlite_saver.asearch(query_2)]
+            assert len(search_results_2) == 1
+            assert search_results_2[0].metadata == self.metadata_2
+
+            search_results_3 = [c async for c in sqlite_saver.asearch(query_3)]
+            assert len(search_results_3) == 2
+
+            search_results_4 = [c async for c in sqlite_saver.asearch(query_4)]
+            assert len(search_results_4) == 0
+
+            # TODO: test before and limit params
diff --git a/tests/checkpoint/test_memory.py b/tests/checkpoint/test_memory.py
new file mode 100644
index 000000000..2a4f26e34
--- /dev/null
+++ b/tests/checkpoint/test_memory.py
@@ -0,0 +1,107 @@
+import pytest
+from langchain_core.runnables import RunnableConfig
+
+from langgraph.checkpoint.base import Checkpoint, CheckpointMetadata
+from langgraph.checkpoint.memory import MemorySaver
+
+
+class TestMemorySaver:
+    @pytest.fixture(autouse=True)
+    def setup(self):
+        self.memory_saver = MemorySaver()
+
+        # objects for test setup
+        self.config_1: RunnableConfig = {
+            "configurable": {"thread_id": "thread-1", "thread_ts": "1"}
+        }
+        self.config_2: RunnableConfig = {
+            "configurable": {"thread_id": "thread-2", "thread_ts": "2"}
+        }
+
+        self.chkpnt_1: Checkpoint = {
+            "v": 1,
+            "ts": "1",
+            "channel_values": {},
+            "channel_versions": {},
+            "versions_seen": {},
+        }
+        self.chkpnt_2: Checkpoint = {
+            "v": 2,
+            "ts": "2",
+            "channel_values": {},
+            "channel_versions": {},
+            "versions_seen": {},
+        }
+
+        self.metadata_1: CheckpointMetadata = {
+            "source": "input",
+            "step": 2,
+            "writes": {},
+            "score": 1,
+        }
+        self.metadata_2: CheckpointMetadata = {
+            "source": "loop",
+            "step": 1,
+            "writes": {"foo": "bar"},
+            "score": None,
+        }
+
+    async def test_search(self):
+        # set up test
+        # save checkpoints
+        self.memory_saver.put(self.config_1, self.chkpnt_1, self.metadata_1)
+        self.memory_saver.put(self.config_2, self.chkpnt_2, self.metadata_2)
+
+        # call method / assertions
+        query_1: CheckpointMetadata = {"source": "input"}  # search by 1 key
+        query_2: CheckpointMetadata = {
+            "step": 1,
+            "writes": {"foo": "bar"},
+        }  # search by multiple keys
+        query_3: CheckpointMetadata = {}  # search by no keys, return all checkpoints
+        query_4: CheckpointMetadata = {"source": "update", "step": 1}  # no match
+
+        search_results_1 = list(self.memory_saver.search(query_1))
+        assert len(search_results_1) == 1
+        assert search_results_1[0].metadata == self.metadata_1
+
+        search_results_2 = list(self.memory_saver.search(query_2))
+        assert len(search_results_2) == 1
+        assert search_results_2[0].metadata == self.metadata_2
+
+        search_results_3 = list(self.memory_saver.search(query_3))
+        assert len(search_results_3) == 2
+
+        search_results_4 = list(self.memory_saver.search(query_4))
+        assert len(search_results_4) == 0
+
+        # TODO: test before and limit params
+
+    async def test_asearch(self):
+        # set up test
+        # save checkpoints
+        self.memory_saver.put(self.config_1, self.chkpnt_1, self.metadata_1)
+        self.memory_saver.put(self.config_2, self.chkpnt_2, self.metadata_2)
+
+        # call method / assertions
+        query_1: CheckpointMetadata = {"source": "input"}  # search by 1 key
+        query_2: CheckpointMetadata = {
+            "step": 1,
+            "writes": {"foo": "bar"},
+        }  # search by multiple keys
+        query_3: CheckpointMetadata = {}  # search by no keys, return all checkpoints
+        query_4: CheckpointMetadata = {"source": "update", "step": 1}  # no match
+
+        search_results_1 = [c async for c in self.memory_saver.asearch(query_1)]
+        assert len(search_results_1) == 1
+        assert search_results_1[0].metadata == self.metadata_1
+
+        search_results_2 = [c async for c in self.memory_saver.asearch(query_2)]
+        assert len(search_results_2) == 1
+        assert search_results_2[0].metadata == self.metadata_2
+
+        search_results_3 = [c async for c in self.memory_saver.asearch(query_3)]
+        assert len(search_results_3) == 2
+
+        search_results_4 = [c async for c in self.memory_saver.asearch(query_4)]
+        assert len(search_results_4) == 0
diff --git a/tests/checkpoint/test_sqlite.py b/tests/checkpoint/test_sqlite.py
new file mode 100644
index 000000000..c64b5ceea
--- /dev/null
+++ b/tests/checkpoint/test_sqlite.py
@@ -0,0 +1,111 @@
+import pytest
+from langchain_core.runnables import RunnableConfig
+
+from langgraph.checkpoint.base import Checkpoint, CheckpointMetadata
+from langgraph.checkpoint.sqlite import SqliteSaver, _metadata_predicate, search_where
+
+
+class TestSqliteSaver:
+    @pytest.fixture(autouse=True)
+    def setup(self):
+        self.sqlite_saver = SqliteSaver.from_conn_string(":memory:")
+
+        # objects for test setup
+        self.config_1: RunnableConfig = {
+            "configurable": {"thread_id": "thread-1", "thread_ts": "1"}
+        }
+        self.config_2: RunnableConfig = {
+            "configurable": {"thread_id": "thread-2", "thread_ts": "2"}
+        }
+
+        self.chkpnt_1: Checkpoint = {
+            "v": 1,
+            "ts": "1",
+            "channel_values": {},
+            "channel_versions": {},
+            "versions_seen": {},
+        }
+        self.chkpnt_2: Checkpoint = {
+            "v": 2,
+            "ts": "2",
+            "channel_values": {},
+            "channel_versions": {},
+            "versions_seen": {},
+        }
+
+        self.metadata_1: CheckpointMetadata = {
+            "source": "input",
+            "step": 2,
+            "writes": {},
+            "score": 1,
+        }
+        self.metadata_2: CheckpointMetadata = {
+            "source": "loop",
+            "step": 1,
+            "writes": {"foo": "bar"},
+            "score": None,
+        }
+        self.metadata_3: CheckpointMetadata = {}
+
+    def test_search(self):
+        # set up test
+        # save checkpoints
+        self.sqlite_saver.put(self.config_1, self.chkpnt_1, self.metadata_1)
+        self.sqlite_saver.put(self.config_2, self.chkpnt_2, self.metadata_2)
+
+        # call method / assertions
+        query_1: CheckpointMetadata = {"source": "input"}  # search by 1 key
+        query_2: CheckpointMetadata = {
+            "step": 1,
+            "writes": {"foo": "bar"},
+        }  # search by multiple keys
+        query_3: CheckpointMetadata = {}  # search by no keys, return all checkpoints
+        query_4: CheckpointMetadata = {"source": "update", "step": 1}  # no match
+
+        search_results_1 = list(self.sqlite_saver.search(query_1))
+        assert len(search_results_1) == 1
+        assert search_results_1[0].metadata == self.metadata_1
+
+        search_results_2 = list(self.sqlite_saver.search(query_2))
+        assert len(search_results_2) == 1
+        assert search_results_2[0].metadata == self.metadata_2
+
+        search_results_3 = list(self.sqlite_saver.search(query_3))
+        assert len(search_results_3) == 2
+
+        search_results_4 = list(self.sqlite_saver.search(query_4))
+        assert len(search_results_4) == 0
+
+        # TODO: test before and limit params
+
+    def test_search_where(self):
+        # call method / assertions
+        expected_predicate_1 = "WHERE json_extract(CAST(metadata AS TEXT), '$.source') = ? AND json_extract(CAST(metadata AS TEXT), '$.step') = ? AND json_extract(CAST(metadata AS TEXT), '$.writes') = ? AND json_extract(CAST(metadata AS TEXT), '$.score') = ? AND thread_ts < ? "
+        expected_param_values_1 = ("input", 2, "{}", 1, "1")
+        assert search_where(self.metadata_1, self.config_1) == (
+            expected_predicate_1,
+            expected_param_values_1,
+        )
+
+    def test_metadata_predicate(self):
+        # call method / assertions
+        expected_predicate_1 = "json_extract(CAST(metadata AS TEXT), '$.source') = ? AND json_extract(CAST(metadata AS TEXT), '$.step') = ? AND json_extract(CAST(metadata AS TEXT), '$.writes') = ? AND json_extract(CAST(metadata AS TEXT), '$.score') = ? "
+        expected_predicate_2 = "json_extract(CAST(metadata AS TEXT), '$.source') = ? AND json_extract(CAST(metadata AS TEXT), '$.step') = ? AND json_extract(CAST(metadata AS TEXT), '$.writes') = ? AND json_extract(CAST(metadata AS TEXT), '$.score') IS ? "
+        expected_predicate_3 = ""
+
+        expected_param_values_1 = ("input", 2, "{}", 1)
+        expected_param_values_2 = ("loop", 1, '{"foo":"bar"}', None)
+        expected_param_values_3 = ()
+
+        assert _metadata_predicate(self.metadata_1) == (
+            expected_predicate_1,
+            expected_param_values_1,
+        )
+        assert _metadata_predicate(self.metadata_2) == (
+            expected_predicate_2,
+            expected_param_values_2,
+        )
+        assert _metadata_predicate(self.metadata_3) == (
+            expected_predicate_3,
+            expected_param_values_3,
+        )
diff --git a/tests/test_pregel.py b/tests/test_pregel.py
index aec36e40e..787946cab 100644
--- a/tests/test_pregel.py
+++ b/tests/test_pregel.py
@@ -5,10 +5,19 @@ import warnings
 from collections import Counter
 from concurrent.futures import ThreadPoolExecutor
 from contextlib import contextmanager
-from typing import Annotated, Any, Generator, Literal, Optional, TypedDict, Union
+from typing import (
+    Annotated,
+    Any,
+    Generator,
+    Literal,
+    Optional,
+    Sequence,
+    TypedDict,
+    Union,
+)
 
 import pytest
-from langchain_core.runnables import RunnableLambda, RunnablePassthrough
+from langchain_core.runnables import RunnableConfig, RunnableLambda, RunnablePassthrough
 from pytest_mock import MockerFixture
 from syrupy import SnapshotAssertion
 
@@ -20,8 +29,9 @@ from langgraph.checkpoint.sqlite import SqliteSaver
 from langgraph.errors import InvalidUpdateError
 from langgraph.graph import END, Graph
 from langgraph.graph.graph import START
-from langgraph.graph.message import MessageGraph
+from langgraph.graph.message import MessageGraph, add_messages
 from langgraph.graph.state import StateGraph
+from langgraph.managed.few_shot import FewShotExamples
 from langgraph.prebuilt.chat_agent_executor import (
     create_function_calling_executor,
     create_tool_calling_executor,
@@ -2676,6 +2686,214 @@ def test_state_graph_w_config(snapshot: SnapshotAssertion) -> None:
     assert app.config_schema().schema_json() == snapshot
 
 
+def test_state_graph_few_shot(snapshot: SnapshotAssertion) -> None:
+    from langchain.chat_models.fake import FakeMessagesListChatModel
+    from langchain_community.tools import tool
+    from langchain_core.messages import AIMessage, AnyMessage, HumanMessage, ToolMessage
+    from langchain_core.prompts import ChatPromptTemplate
+
+    class BaseState(TypedDict):
+        messages: Annotated[list[AnyMessage], add_messages]
+
+    class AgentState(BaseState):
+        examples: Annotated[
+            Sequence[BaseState], FewShotExamples[BaseState].configure(k=1)
+        ]
+
+    # Assemble the tools
+    @tool()
+    def search_api(query: str) -> str:
+        """Searches the API for the query."""
+        return f"result for {query}"
+
+    tools = [search_api]
+
+    prompt = ChatPromptTemplate.from_messages(
+        [
+            (
+                "system",
+                """You are a nice assistant.
+Some examples of past conversations:
+{examples}""",
+            ),
+            ("placeholder", "{messages}"),
+        ]
+    )
+
+    model = FakeMessagesListChatModel(
+        responses=[
+            AIMessage(
+                content="",
+                tool_calls=[
+                    {
+                        "id": "tool_call123",
+                        "name": "search_api",
+                        "args": {"query": "query"},
+                    },
+                ],
+            ),
+            AIMessage(content="answer"),
+        ]
+    )
+
+    def agent(state: AgentState, config: RunnableConfig) -> AgentState:
+        # begin: testing code
+        assert state["examples"] == config["configurable"]["expected_examples"]
+        # end: testing code
+        formatted = prompt.invoke(state)
+        response = model.invoke(formatted)
+        return {"messages": response}
+
+    # Define decision-making logic
+    def should_continue(data: AgentState) -> str:
+        # Logic to decide whether to continue in the loop or exit
+        if not data["messages"][-1].tool_calls:
+            return "exit"
+        else:
+            return "continue"
+
+    # Define a new graph
+    workflow = StateGraph(AgentState)
+
+    workflow.add_node("agent", agent)
+    workflow.add_node("tools", ToolNode(tools))
+    workflow.set_entry_point("agent")
+    workflow.add_conditional_edges(
+        "agent", should_continue, {"continue": "tools", "exit": END}
+    )
+    workflow.add_edge("tools", "agent")
+
+    with SqliteSaver.from_conn_string(":memory:") as saver:
+        app = workflow.compile(checkpointer=saver)
+
+        first_messages = [
+            HumanMessage(content="what is weather in sf", id=AnyStr()),
+            AIMessage(
+                content="",
+                id=AnyStr(),
+                tool_calls=[
+                    {
+                        "name": "search_api",
+                        "args": {"query": "query"},
+                        "id": "tool_call123",
+                    }
+                ],
+            ),
+            ToolMessage(
+                content="result for query",
+                name="search_api",
+                id=AnyStr(),
+                tool_call_id="tool_call123",
+            ),
+            AIMessage(content="answer", id=AnyStr()),
+        ]
+        assert app.invoke(
+            {"messages": "what is weather in sf"},
+            {"configurable": {"thread_id": "1", "expected_examples": []}},
+        ) == {"messages": first_messages}
+
+        # get first checkpoint
+        chkpnt_tuple_1 = saver.get_tuple({"configurable": {"thread_id": "1"}})
+        config = chkpnt_tuple_1.config
+        checkpoint = chkpnt_tuple_1.checkpoint
+        metadata = chkpnt_tuple_1.metadata
+
+        # not needed in application code, only for testing
+        hiscored = list(saver.search({"score": 1}))
+        assert hiscored == []
+
+        # mark as "good"
+        metadata["score"] = 1
+        saver.put(config, checkpoint, metadata)
+
+        # not needed in application code, only for testing
+        hiscored = list(saver.search({"score": 1}))
+        assert len(hiscored) == 1
+        assert hiscored[0].checkpoint["channel_values"]["messages"] == first_messages
+
+        second_messages = [
+            HumanMessage(content="what is weather in la", id=AnyStr()),
+            AIMessage(
+                content="",
+                id=AnyStr(),
+                tool_calls=[
+                    {
+                        "name": "search_api",
+                        "args": {"query": "query"},
+                        "id": "tool_call123",
+                    }
+                ],
+            ),
+            ToolMessage(
+                content="result for query",
+                name="search_api",
+                id=AnyStr(),
+                tool_call_id="tool_call123",
+            ),
+            AIMessage(content="answer", id=AnyStr()),
+        ]
+        assert app.invoke(
+            {"messages": "what is weather in la"},
+            {
+                "configurable": {
+                    "thread_id": "2",
+                    # below is only for testing purposes, not part of few shot api
+                    "expected_examples": [{"messages": first_messages}],
+                }
+            },
+        ) == {"messages": second_messages}
+
+        # get first checkpoint
+        chkpnt_tuple_2 = saver.get_tuple({"configurable": {"thread_id": "2"}})
+        config = chkpnt_tuple_2.config
+        checkpoint = chkpnt_tuple_2.checkpoint
+        metadata = chkpnt_tuple_2.metadata
+
+        # not needed in application code, only for testing
+        hiscored = list(saver.search({"score": 1}))
+        assert len(hiscored) == 1
+
+        # mark as "good"
+        metadata["score"] = 1
+        saver.put(config, checkpoint, metadata)
+
+        hiscored = list(saver.search({"score": 1}))
+        assert len(hiscored) == 2
+
+        assert app.invoke(
+            {"messages": "what is weather in ny"},
+            {
+                "configurable": {
+                    "thread_id": "3",
+                    # below is only for testing purposes, not part of few shot api
+                    "expected_examples": [{"messages": second_messages}],
+                }
+            },
+        ) == {
+            "messages": [
+                HumanMessage(content="what is weather in ny", id=AnyStr()),
+                AIMessage(
+                    content="",
+                    id=AnyStr(),
+                    tool_calls=[
+                        {
+                            "name": "search_api",
+                            "args": {"query": "query"},
+                            "id": "tool_call123",
+                        }
+                    ],
+                ),
+                ToolMessage(
+                    content="result for query",
+                    name="search_api",
+                    id=AnyStr(),
+                    tool_call_id="tool_call123",
+                ),
+                AIMessage(content="answer", id=AnyStr()),
+            ]
+        }
+
+
 def test_conditional_entrypoint_graph_state(snapshot: SnapshotAssertion) -> None:
     class AgentState(TypedDict, total=False):
         input: str
diff --git a/tests/test_pregel_async.py b/tests/test_pregel_async.py
index 1b78c6f95..f6f0bb50b 100644
--- a/tests/test_pregel_async.py
+++ b/tests/test_pregel_async.py
@@ -10,6 +10,7 @@ from typing import (
     AsyncIterator,
     Generator,
     Optional,
+    Sequence,
     TypedDict,
     Union,
 )
@@ -28,12 +29,14 @@ from langgraph.checkpoint.aiosqlite import AsyncSqliteSaver
 from langgraph.errors import InvalidUpdateError
 from langgraph.graph import END, Graph, StateGraph
 from langgraph.graph.graph import START
-from langgraph.graph.message import MessageGraph
+from langgraph.graph.message import MessageGraph, add_messages
+from langgraph.managed.few_shot import FewShotExamples
 from langgraph.prebuilt.chat_agent_executor import (
     create_function_calling_executor,
     create_tool_calling_executor,
 )
 from langgraph.prebuilt.tool_executor import ToolExecutor
+from langgraph.prebuilt.tool_node import ToolNode
 from langgraph.pregel import Channel, GraphRecursionError, Pregel, StateSnapshot
 from tests.any_str import AnyStr
 from tests.memory_assert import MemorySaverAssertImmutable
@@ -2415,6 +2418,162 @@ async def test_conditional_graph_state() -> None:
     )
 
 
+async def test_state_graph_few_shot() -> None:
+    from langchain.chat_models.fake import FakeMessagesListChatModel
+    from langchain_community.tools import tool
+    from langchain_core.messages import AIMessage, AnyMessage, HumanMessage, ToolMessage
+    from langchain_core.prompts import ChatPromptTemplate
+
+    class BaseState(TypedDict):
+        messages: Annotated[list[AnyMessage], add_messages]
+
+    class AgentState(BaseState):
+        examples: Annotated[Sequence[BaseState], FewShotExamples[BaseState]]
+
+    # Assemble the tools
+    @tool()
+    def search_api(query: str) -> str:
+        """Searches the API for the query."""
+        return f"result for {query}"
+
+    tools = [search_api]
+
+    prompt = ChatPromptTemplate.from_messages(
+        [
+            (
+                "system",
+                """You are a nice assistant.
+Some examples of past conversations:
+{examples}""",
+            ),
+            ("placeholder", "{messages}"),
+        ]
+    )
+
+    model = FakeMessagesListChatModel(
+        responses=[
+            AIMessage(
+                content="",
+                tool_calls=[
+                    {
+                        "id": "tool_call123",
+                        "name": "search_api",
+                        "args": {"query": "query"},
+                    },
+                ],
+            ),
+            AIMessage(content="answer"),
+        ]
+    )
+
+    async def agent(state: AgentState, config: RunnableConfig) -> AgentState:
+        # begin: testing code
+        assert state["examples"] == config["configurable"]["expected_examples"]
+        # end: testing code
+        formatted = await prompt.ainvoke(state)
+        response = await model.ainvoke(formatted)
+        return {"messages": response}
+
+    # Define decision-making logic
+    def should_continue(data: AgentState) -> str:
+        # Logic to decide whether to continue in the loop or exit
+        if not data["messages"][-1].tool_calls:
+            return "exit"
+        else:
+            return "continue"
+
+    # Define a new graph
+    workflow = StateGraph(AgentState)
+
+    workflow.add_node("agent", agent)
+    workflow.add_node("tools", ToolNode(tools))
+    workflow.set_entry_point("agent")
+    workflow.add_conditional_edges(
+        "agent", should_continue, {"continue": "tools", "exit": END}
+    )
+    workflow.add_edge("tools", "agent")
+
+    async with AsyncSqliteSaver.from_conn_string(":memory:") as saver:
+        app = workflow.compile(checkpointer=saver)
+
+        first_messages = [
+            HumanMessage(content="what is weather in sf", id=AnyStr()),
+            AIMessage(
+                content="",
+                id=AnyStr(),
+                tool_calls=[
+                    {
+                        "name": "search_api",
+                        "args": {"query": "query"},
+                        "id": "tool_call123",
+                    }
+                ],
+            ),
+            ToolMessage(
+                content="result for query",
+                name="search_api",
+                id=AnyStr(),
+                tool_call_id="tool_call123",
+            ),
+            AIMessage(content="answer", id=AnyStr()),
+        ]
+        assert await app.ainvoke(
+            {"messages": "what is weather in sf"},
+            {"configurable": {"thread_id": "1", "expected_examples": []}},
+        ) == {"messages": first_messages}
+
+        # get first checkpoint
+        chkpnt_tuple_1 = await saver.aget_tuple({"configurable": {"thread_id": "1"}})
+        config = chkpnt_tuple_1.config
+        checkpoint = chkpnt_tuple_1.checkpoint
+        metadata = chkpnt_tuple_1.metadata
+
+        # not needed in application code, only for testing
+        assert [c async for c in saver.asearch({"score": 1})] == []
+
+        # mark as "good"
+        metadata["score"] = 1
+        await saver.aput(config, checkpoint, metadata)
+
+        # not needed in application code, only for testing
+        hiscored = [c async for c in saver.asearch({"score": 1})]
+        assert len(hiscored) == 1
+        assert hiscored[0].checkpoint["channel_values"]["messages"] == first_messages
+
+        assert await app.ainvoke(
+            {"messages": "what is weather in la"},
+            {
+                "configurable": {
+                    "thread_id": "2",
+                    # below is only for testing purposes, not part of few shot api
+                    "expected_examples": [{"messages": first_messages}],
+                }
+            },
+        ) == {
+            "messages": [
+                HumanMessage(content="what is weather in la", id=AnyStr()),
+                AIMessage(
+                    content="",
+                    id=AnyStr(),
+                    tool_calls=[
+                        {
+                            "name": "search_api",
+                            "args": {"query": "query"},
+                            "id": "tool_call123",
+                        }
+                    ],
+                ),
+                ToolMessage(
+                    content="result for query",
+                    name="search_api",
+                    id=AnyStr(),
+                    tool_call_id="tool_call123",
+                ),
+                AIMessage(content="answer", id=AnyStr()),
+            ]
+        }
+
+
 async def test_conditional_entrypoint_graph() -> None:
     async def left(data: str) -> str:
         return data + "->left"