Add Workflow Executor Example (#892)

Signed-off-by: JoshuaL3000 <joshua.jian.ern.liew@intel.com> Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
2024-11-01 09:50:20 +08:00
parent c65d7d40fb
commit bf5c391e47
16 changed files with 731 additions and 0 deletions
--- a/WorkflowExecAgent/README.md
+++ b/WorkflowExecAgent/README.md
@@ -0,0 +1,131 @@
+# Workflow Executor Agent
+
+## Overview
+
+GenAI Workflow Executor Example showcases the capability to handle data/AI workflow operations via LangChain agents to execute custom-defined workflow-based tools. These workflow tools can be interfaced from any 3rd-party tools in the market (no-code/low-code/IDE) such as Alteryx, RapidMiner, Power BI, Intel Data Insight Automation which allows users to create complex data/AI workflow operations for different use-cases.
+
+### Workflow Executor
+
+This example demonstrates a single React-LangGraph with a `Workflow Executor` tool to ingest a user prompt to execute workflows and return an agent reasoning response based on the workflow output data.
+
+First the LLM extracts the relevant information from the user query based on the schema of the tool in `tools/tools.yaml`. Then the agent sends this `AgentState` to the `Workflow Executor` tool.
+
+`Workflow Executor` tool uses `EasyDataSDK` class as seen under `tools/sdk.py` to interface with several high-level API's. There are 3 steps to this tool implementation:
+
+1. Starts the workflow with workflow parameters and workflow id extracted from the user query.
+
+2. Periodically checks the workflow status for completion or failure. This may be through a database which stores the current status of the workflow
+
+3. Retrieves the output data from the workflow through a storage service.
+
+The `AgentState` is sent back to the LLM for reasoning. Based on the output data, the LLM generates a response to answer the user's input prompt.
+
+Below shows an illustration of this flow:
+
+![image](https://github.com/user-attachments/assets/cb135042-1505-4aef-8822-c78c2f72aa2a)
+
+### Workflow Serving for Agent
+
+As an example, here we have a Churn Prediction use-case workflow as the serving workflow for the agent execution. It is created through Intel Data Insight Automation platform. The image below shows a snapshot of the Churn Prediction workflow.
+
+![image](https://github.com/user-attachments/assets/c067f8b3-86cf-4abc-a8bd-51a98de8172d)
+
+The workflow contains 2 paths which can be seen in the workflow illustrated, the top path and bottom path.
+
+1. Top path - The training path which ends at the random forest classifier node is the training path. The data is cleaned through a series of nodes and used to train a random forest model for prediction.
+
+2. Bottom path - The inference path where trained random forest model is used for inferencing based on input parameter.
+
+For this agent workflow execution, the inferencing path is executed to yield the final output result of the `Model Predictor` node. The same output is returned to the `Workflow Executor` tool through the `Langchain API Serving` node.
+
+There are `Serving Parameters` in the workflow, which are the tool input variables used to start a workflow instance obtained from `params` the LLM extracts from the user query. Below shows the parameter configuration option for the Intel Data Insight Automation workflow UI.
+
+![image](https://github.com/user-attachments/assets/ce8ef01a-56ff-4278-b84d-b6e4592b28c6)
+
+Manually running the workflow yields the tabular data output as shown below:
+
+![image](https://github.com/user-attachments/assets/241c1aba-2a24-48da-8005-ec7bfe657179)
+
+In the workflow serving for agent, this output will be returned to the `Workflow Executor` tool. The LLM can then answer the user's original question based on this output.
+
+To start prompting the agent microservice, we will use the following command for this use case:
+
+```sh
+curl http://${ip_address}:9090/v1/chat/completions -X POST -H "Content-Type: application/json" -d '{
+     "query": "I have a data with gender Female, tenure 55, MonthlyAvgCharges 103.7. Predict if this entry will churn. My workflow id is '${workflow_id}'."
+    }'
+```
+
+The user has to provide a `workflow_id` and workflow `params` in the query. `workflow_id` a unique id used for serving the workflow to the microservice. Notice that the `query` string includes all the workflow `params` which the user has defined in the workflow. The LLM will extract these parameters into a dictionary format for the workflow `Serving Parameters` as shown below:
+
+```python
+params = {"gender": "Female", "tenure": 55, "MonthlyAvgCharges": 103.7}
+```
+
+These parameters will be passed into the `Workflow Executor` tool to start the workflow execution of specified `workflow_id`. Thus, everything will be handled via the microservice.
+
+And finally here are the results from the microservice logs:
+
+![image](https://github.com/user-attachments/assets/969fefb7-543d-427f-a56c-dc70e474ae60)
+
+## Microservice Setup
+
+### Start Agent Microservice
+
+Workflow Executor will have a single docker image. First, build the agent docker image.
+
+```sh
+git clone https://github.com/opea-project/GenAIExamples.git
+cd GenAIExamples//WorkflowExecAgent/docker_image_build/
+docker compose -f build.yaml build --no-cache
+```
+
+Configure `GenAIExamples/WorkflowExecAgent/docker_compose/.env` file with the following. Replace the variables according to your usecase.
+
+```sh
+export SDK_BASE_URL=${SDK_BASE_URL}
+export SERVING_TOKEN=${SERVING_TOKEN}
+export HUGGINGFACEHUB_API_TOKEN=${HF_TOKEN}
+export llm_engine=${llm_engine}
+export llm_endpoint_url=${llm_endpoint_url}
+export ip_address=$(hostname -I | awk '{print $1}')
+export model="mistralai/Mistral-7B-Instruct-v0.3"
+export recursion_limit=${recursion_limit}
+export temperature=0
+export max_new_tokens=1000
+export WORKDIR=${WORKDIR}
+export TOOLSET_PATH=$WORKDIR/GenAIExamples/WorkflowExecAgent/tools/
+export http_proxy=${http_proxy}
+export https_proxy=${https_proxy}
+```
+
+Launch service by running the docker compose command.
+
+```sh
+cd $WORKDIR/GenAIExamples/WorkflowExecAgent/docker_compose
+docker compose -f compose.yaml up -d
+```
+
+### Validate service
+
+The microservice logs can be viewed using:
+
+```sh
+docker logs workflowexec-agent-endpoint
+```
+
+You should be able to see "HTTP server setup successful" upon successful startup.
+
+You can validate the service using the following command:
+
+```sh
+curl http://${ip_address}:9090/v1/chat/completions -X POST -H "Content-Type: application/json" -d '{
+     "query": "I have a data with gender Female, tenure 55, MonthlyAvgCharges 103.7. Predict if this entry will churn. My workflow id is '${workflow_id}'."
+    }'
+```
+
+Update the `query` with the workflow parameters, workflow id, etc based on the workflow context.
+
+## Roadmap
+
+Phase II: Agent memory integration to enable capability to store tool intermediate results, such as workflow instance key.
--- a/WorkflowExecAgent/docker_compose/intel/cpu/xeon/compose_vllm.yaml
+++ b/WorkflowExecAgent/docker_compose/intel/cpu/xeon/compose_vllm.yaml
@@ -0,0 +1,31 @@
+# Copyright (C) 2024 Intel Corporation
+# SPDX-License-Identifier: Apache-2.0
+
+services:
+  worflowexec-agent:
+    image: opea/agent-langchain:latest
+    container_name: workflowexec-agent-endpoint
+    volumes:
+      - ${WORKDIR}/GenAIComps/comps/agent/langchain/:/home/user/comps/agent/langchain/
+      - ${TOOLSET_PATH}:/home/user/tools/
+    ports:
+      - "9090:9090"
+    ipc: host
+    environment:
+      ip_address: ${ip_address}
+      strategy: react_langgraph
+      recursion_limit: ${recursion_limit}
+      llm_engine: ${llm_engine}
+      llm_endpoint_url: ${llm_endpoint_url}
+      model: ${model}
+      temperature: ${temperature}
+      max_new_tokens: ${max_new_tokens}
+      streaming: false
+      tools: /home/user/tools/tools.yaml
+      no_proxy: ${no_proxy}
+      http_proxy: ${http_proxy}
+      https_proxy: ${https_proxy}
+      port: 9090
+      SDK_BASE_URL: ${SDK_BASE_URL}
+      SERVING_TOKEN: ${SERVING_TOKEN}
+      custom_prompt: /home/user/tools/custom_prompt.py
--- a/WorkflowExecAgent/docker_image_build/build.yaml
+++ b/WorkflowExecAgent/docker_image_build/build.yaml
@@ -0,0 +1,13 @@
+# Copyright (C) 2024 Intel Corporation
+# SPDX-License-Identifier: Apache-2.0
+
+services:
+  agent-langchain:
+    build:
+      context: GenAIComps
+      dockerfile: comps/agent/langchain/Dockerfile
+      args:
+        http_proxy: ${http_proxy}
+        https_proxy: ${https_proxy}
+        no_proxy: ${no_proxy}
+    image: ${REGISTRY:-opea}/agent-langchain:${TAG:-latest}
--- a/WorkflowExecAgent/tests/1_build_images.sh
+++ b/WorkflowExecAgent/tests/1_build_images.sh
@@ -0,0 +1,29 @@
+#!/bin/bash
+# Copyright (C) 2024 Intel Corporation
+# SPDX-License-Identifier: Apache-2.0
+
+set -e
+WORKPATH=$(dirname "$PWD")
+export WORKDIR=$WORKPATH/../../
+echo "WORKDIR=${WORKDIR}"
+
+function get_genai_comps() {
+    if [ ! -d "GenAIComps" ] ; then
+        git clone https://github.com/opea-project/GenAIComps.git && cd GenAIComps && git checkout "${opea_branch:-"main"}" && cd ../
+    fi
+}
+
+function build_agent_docker_image() {
+    cd $WORKDIR/GenAIExamples/WorkflowExecAgent/docker_image_build/
+    get_genai_comps
+    echo "Build agent image with --no-cache..."
+    docker compose -f build.yaml build --no-cache
+}
+
+function main() {
+    echo "==================== Build agent docker image ===================="
+    build_agent_docker_image
+    echo "==================== Build agent docker image completed ===================="
+}
+
+main
--- a/WorkflowExecAgent/tests/2_start_vllm_service.sh
+++ b/WorkflowExecAgent/tests/2_start_vllm_service.sh
@@ -0,0 +1,67 @@
+#!/bin/bash
+# Copyright (C) 2024 Intel Corporation
+# SPDX-License-Identifier: Apache-2.0
+
+set -e
+
+WORKPATH=$(dirname "$PWD")
+LOG_PATH="$WORKPATH/tests"
+vllm_port=${vllm_port}
+[[ -z "$vllm_port" ]] && vllm_port=8084
+model=mistralai/Mistral-7B-Instruct-v0.3
+export WORKDIR=$WORKPATH/../../
+export HF_TOKEN=${HUGGINGFACEHUB_API_TOKEN}
+
+function build_vllm_docker_image() {
+    echo "Building the vllm docker images"
+    cd $WORKPATH
+    echo $WORKPATH
+    if [ ! -d "./vllm" ]; then
+        git clone https://github.com/vllm-project/vllm.git
+        cd ./vllm; git checkout tags/v0.6.0
+    else
+        cd ./vllm
+    fi
+    docker build -f Dockerfile.cpu -t vllm-cpu-env --shm-size=100g .
+    if [ $? -ne 0 ]; then
+        echo "opea/vllm:cpu failed"
+        exit 1
+    else
+        echo "opea/vllm:cpu successful"
+    fi
+}
+
+function start_vllm_service() {
+    echo "start vllm service"
+    docker run -d -p ${vllm_port}:${vllm_port} --rm --network=host --name test-comps-vllm-service -v ~/.cache/huggingface:/root/.cache/huggingface -v ${WORKPATH}/tests/tool_chat_template_mistral_custom.jinja:/root/tool_chat_template_mistral_custom.jinja -e HF_TOKEN=$HF_TOKEN -e http_proxy=$http_proxy -e https_proxy=$https_proxy -it vllm-cpu-env --model ${model} --port ${vllm_port} --chat-template /root/tool_chat_template_mistral_custom.jinja --enable-auto-tool-choice --tool-call-parser mistral
+    echo ${LOG_PATH}/vllm-service.log
+    sleep 5s
+    echo "Waiting vllm ready"
+    n=0
+    until [[ "$n" -ge 100 ]] || [[ $ready == true ]]; do
+        docker logs test-comps-vllm-service &> ${LOG_PATH}/vllm-service.log
+        n=$((n+1))
+        if grep -q "Uvicorn running on" ${LOG_PATH}/vllm-service.log; then
+            break
+        fi
+        if grep -q "No such container" ${LOG_PATH}/vllm-service.log; then
+            echo "container test-comps-vllm-service not found"
+            exit 1
+        fi
+        sleep 5s
+    done
+    sleep 5s
+    echo "Service started successfully"
+}
+
+function main() {
+    echo "==================== Build vllm docker image ===================="
+    build_vllm_docker_image
+    echo "==================== Build vllm docker image completed ===================="
+
+    echo "==================== Start vllm docker service ===================="
+    start_vllm_service
+    echo "==================== Start vllm docker service completed ===================="
+}
+
+main
--- a/WorkflowExecAgent/tests/3_launch_and_validate_agent.sh
+++ b/WorkflowExecAgent/tests/3_launch_and_validate_agent.sh
@@ -0,0 +1,66 @@
+#!/bin/bash
+# Copyright (C) 2024 Intel Corporation
+# SPDX-License-Identifier: Apache-2.0
+
+set -e
+
+WORKPATH=$(dirname "$PWD")
+workflow_id=9809
+vllm_port=${vllm_port}
+[[ -z "$vllm_port" ]] && vllm_port=8084
+export WORKDIR=$WORKPATH/../../
+echo "WORKDIR=${WORKDIR}"
+export SDK_BASE_URL=${SDK_BASE_URL}
+export SERVING_TOKEN=${SERVING_TOKEN}
+export HF_TOKEN=${HUGGINGFACEHUB_API_TOKEN}
+export llm_engine=vllm
+export ip_address=$(hostname -I | awk '{print $1}')
+export llm_endpoint_url=http://${ip_address}:${vllm_port}
+export model=mistralai/Mistral-7B-Instruct-v0.3
+export recursion_limit=25
+export temperature=0
+export max_new_tokens=1000
+export TOOLSET_PATH=$WORKDIR/GenAIExamples/WorkflowExecAgent/tools/
+
+function start_agent_and_api_server() {
+    echo "Starting Agent services"
+    cd $WORKDIR/GenAIExamples/WorkflowExecAgent/docker_compose/intel/cpu/xeon
+    WORKDIR=$WORKPATH/docker_image_build/ docker compose -f compose_vllm.yaml up -d
+    echo "Waiting agent service ready"
+    sleep 5s
+}
+
+function validate() {
+    local CONTENT="$1"
+    local EXPECTED_RESULT="$2"
+    local SERVICE_NAME="$3"
+
+    if echo "$CONTENT" | grep -q "$EXPECTED_RESULT"; then
+        echo "[ $SERVICE_NAME ] Content is as expected: $CONTENT"
+        echo "[TEST INFO]: Workflow Executor agent service PASSED"
+    else
+        echo "[ $SERVICE_NAME ] Content does not match the expected result: $CONTENT"
+        echo "[TEST INFO]: Workflow Executor agent service FAILED"
+    fi
+}
+
+function validate_agent_service() {
+    echo "----------------Test agent ----------------"
+    local CONTENT=$(curl http://${ip_address}:9090/v1/chat/completions -X POST -H "Content-Type: application/json" -d '{
+     "query": "I have a data with gender Female, tenure 55, MonthlyAvgCharges 103.7. Predict if this entry will churn. My workflow id is '${workflow_id}'."
+    }')
+    validate "$CONTENT" "The entry is not likely to churn" "workflowexec-agent-endpoint"
+    docker logs workflowexec-agent-endpoint
+}
+
+function main() {
+    echo "==================== Start agent ===================="
+    start_agent_and_api_server
+    echo "==================== Agent started ===================="
+
+    echo "==================== Validate agent service ===================="
+    validate_agent_service
+    echo "==================== Agent service validated ===================="
+}
+
+main
--- a/WorkflowExecAgent/tests/README.md
+++ b/WorkflowExecAgent/tests/README.md
@@ -0,0 +1,38 @@
+# Validate Workflow Agent Microservice
+
+Microservice validation for Intel Data Insight Automation platform workflow serving.
+
+## Usage
+
+Configure necessary variables as listed below. Replace the variables according to your usecase.
+
+```sh
+export SDK_BASE_URL=${SDK_BASE_URL}
+export SERVING_TOKEN=${SERVING_TOKEN}
+export HUGGINGFACEHUB_API_TOKEN=${HF_TOKEN}
+export workflow_id=${workflow_id}       # workflow_id of the serving workflow
+export vllm_port=${vllm_port}           # vllm serving port
+export ip_address=$(hostname -I | awk '{print $1}')
+export VLLM_CPU_OMP_THREADS_BIND=${VLLM_CPU_OMP_THREADS_BIND}
+export http_proxy=${http_proxy}
+export https_proxy=${https_proxy}
+```
+
+Note: `SDK_BASE_URL` and `SERVING_TOKEN` can be obtained from Intel Data Insight Automation platform.
+
+Launch validation by running the following command.
+
+```sh
+cd GenAIExamples/WorkflowExecAgent/tests
+. /test_compose_on_xeon.sh
+```
+
+`test_compose_on_xeon.sh` will run the other `.sh` files under `tests/`. The validation script launches 1 docker container for the agent microservice, and another for the vllm model serving on CPU. When validation is completed, all containers will be stopped.
+
+The validation is tested by checking if the model reasoning output response matches a partial substring. The expected output is shown below:
+
+![image](https://github.com/user-attachments/assets/88081bc8-7b73-470d-970e-92e0fe5f96ec)
+
+## Note
+
+- Currently the validation test is only designed with vllm model serving (CPU only).
--- a/WorkflowExecAgent/tests/test_compose_vllm_on_xeon.sh
+++ b/WorkflowExecAgent/tests/test_compose_vllm_on_xeon.sh
@@ -0,0 +1,33 @@
+# Copyright (C) 2024 Intel Corporation
+# SPDX-License-Identifier: Apache-2.0
+
+function stop_agent_and_api_server() {
+    echo "Stopping Agent services"
+    docker rm --force $(docker ps -a -q --filter="name=workflowexec-agent-endpoint")
+}
+
+function stop_vllm_docker() {
+    cid=$(docker ps -aq --filter "name=test-comps-vllm-service")
+    echo "Stopping the docker containers "${cid}
+    if [[ ! -z "$cid" ]]; then docker rm $cid -f && sleep 1s; fi
+    echo "Docker containers stopped successfully"
+}
+
+echo "=================== #1 Building docker images ===================="
+bash 1_build_images.sh
+echo "=================== #1 Building docker images completed ===================="
+
+echo "=================== #2 Start vllm service ===================="
+bash 2_start_vllm_service.sh
+echo "=================== #2 Start vllm service completed ===================="
+
+echo "=================== #3 Start agent and API server ===================="
+bash 3_launch_and_validate_agent.sh
+echo "=================== #3 Agent test completed ===================="
+
+echo "=================== #4 Stop agent and API server ===================="
+stop_agent_and_api_server
+stop_vllm_docker
+echo "=================== #4 Agent and API server stopped ===================="
+
+echo "ALL DONE!"
--- a/WorkflowExecAgent/tests/tool_chat_template_mistral_custom.jinja
+++ b/WorkflowExecAgent/tests/tool_chat_template_mistral_custom.jinja
@@ -0,0 +1,83 @@
+{%- if messages[0]["role"] == "system" %}
+    {%- set system_message = messages[0]["content"] %}
+    {%- set loop_messages = messages[1:] %}
+{%- else %}
+    {%- set loop_messages = messages %}
+{%- endif %}
+{%- if not tools is defined %}
+    {%- set tools = none %}
+{%- endif %}
+{%- set user_messages = loop_messages | selectattr("role", "equalto", "user") | list %}
+
+{%- for message in loop_messages | rejectattr("role", "equalto", "tool") | rejectattr("role", "equalto", "tool_results") | selectattr("tool_calls", "undefined") %}
+{%- endfor %}
+
+{{- bos_token }}
+{%- for message in loop_messages %}
+    {%- if message["role"] == "user" %}
+        {%- if tools is not none and (message == user_messages[-1]) %}
+            {{- "[AVAILABLE_TOOLS] [" }}
+            {%- for tool in tools %}
+                {%- set tool = tool.function %}
+                {{- '{"type": "function", "function": {' }}
+                {%- for key, val in tool.items() if key != "return" %}
+                    {%- if val is string %}
+                        {{- '"' + key + '": "' + val + '"' }}
+                    {%- else %}
+                        {{- '"' + key + '": ' + val|tojson }}
+                    {%- endif %}
+                    {%- if not loop.last %}
+                        {{- ", " }}
+                    {%- endif %}
+                {%- endfor %}
+                {{- "}}" }}
+                {%- if not loop.last %}
+                    {{- ", " }}
+                {%- else %}
+                    {{- "]" }}
+                {%- endif %}
+            {%- endfor %}
+            {{- "[/AVAILABLE_TOOLS]" }}
+        {%- endif %}
+        {%- if loop.last and system_message is defined %}
+            {{- "[INST] " + system_message + "\n\n" + message["content"] + "[/INST]" }}
+        {%- else %}
+            {{- "[INST] " + message["content"] + "[/INST]" }}
+        {%- endif %}
+    {%- elif message["role"] == "tool_calls" or message.tool_calls is defined %}
+        {%- if message.tool_calls is defined %}
+            {%- set tool_calls = message.tool_calls %}
+        {%- else %}
+            {%- set tool_calls = message.content %}
+        {%- endif %}
+        {{- "[TOOL_CALLS] [" }}
+        {%- for tool_call in tool_calls %}
+            {%- set out = tool_call.function|tojson %}
+            {{- out[:-1] }}
+            {%- if not tool_call.id is defined or tool_call.id|length < 9 %}
+                {{- raise_exception("Tool call IDs should be alphanumeric strings with length >= 9! (1)" + tool_call.id) }}
+            {%- endif %}
+            {{- ', "id": "' + tool_call.id[-9:] + '"}' }}
+            {%- if not loop.last %}
+                {{- ", " }}
+            {%- else %}
+                {{- "]" + eos_token }}
+            {%- endif %}
+        {%- endfor %}
+    {%- elif message["role"] == "assistant" %}
+        {{- " " + message["content"] + eos_token }}
+    {%- elif message["role"] == "tool_results" or message["role"] == "tool" %}
+        {%- if message.content is defined and message.content.content is defined %}
+            {%- set content = message.content.content %}
+        {%- else %}
+            {%- set content = message.content %}
+        {%- endif %}
+        {{- '[TOOL_RESULTS] {"content": ' + content|string + ", " }}
+        {%- if not message.tool_call_id is defined or message.tool_call_id|length < 9 %}
+            {{- raise_exception("Tool call IDs should be alphanumeric strings with length >= 9! (2)" + message.tool_call_id) }}
+        {%- endif %}
+        {{- '"call_id": "' + message.tool_call_id[-9:] + '"}[/TOOL_RESULTS]' }}
+    {%- else %}
+        {{- raise_exception("Only user and assistant roles are supported, with the exception of an initial optional system message!") }}
+    {%- endif %}
+{%- endfor %}
--- a/WorkflowExecAgent/tools/components/component.py
+++ b/WorkflowExecAgent/tools/components/component.py
@@ -0,0 +1,10 @@
+# Copyright (C) 2024 Intel Corporation
+# SPDX-License-Identifier: Apache-2.0
+
+
+class Component:
+    def __init__(self, request_handler):
+        self.request_handler = request_handler
+
+    def _make_request(self, *args, **kwargs):
+        return self.request_handler._make_request(*args, **kwargs)
--- a/WorkflowExecAgent/tools/components/workflow.py
+++ b/WorkflowExecAgent/tools/components/workflow.py
@@ -0,0 +1,70 @@
+# Copyright (C) 2024 Intel Corporation
+# SPDX-License-Identifier: Apache-2.0
+
+import json
+from typing import Dict
+
+from tools.components.component import Component
+
+
+class Workflow(Component):
+    """Class for handling EasyData workflow operations.
+
+    Attributes:
+        workflow_id: workflow id
+        wf_key: workflow key. Generated and stored when starting a servable workflow.
+    """
+
+    def __init__(self, request_handler, workflow_id=None, workflow_key=None):
+        super().__init__(request_handler)
+        self.workflow_id = workflow_id
+        self.wf_key = workflow_key
+
+    def start(self, params: Dict[str, str]) -> Dict[str, str]:
+        """
+        ``POST https://SDK_BASE_URL/serving/servable_workflows/{workflow_id}/start``
+
+        Starts a workflow instance with the workflow_id and parameters provided.
+        Returns a workflow key used to track the workflow instance.
+
+        :param dict params: Workflow parameters used to start workflow.
+
+        :returns: WorkflowKey
+
+        :rtype: string
+        """
+        data = json.dumps({"params": params})
+        endpoint = f"serving/servable_workflows/{self.workflow_id}/start"
+        self.wf_key = self._make_request(endpoint, "POST", data)["wf_key"]
+        if self.wf_key:
+            return f"Workflow successfully started. The workflow key is {self.wf_key}."
+        else:
+            return "Workflow failed to start"
+
+    def get_status(self) -> Dict[str, str]:
+        """
+        ``GET https://SDK_BASE_URL/serving/serving_workflows/{workflow_key}/status``
+
+        Gets the workflow status.
+
+        :returns: WorkflowStatus
+
+        :rtype: string
+        """
+
+        endpoint = f"serving/serving_workflows/{self.wf_key}/status"
+        return self._make_request(endpoint, "GET")
+
+    def result(self) -> list[Dict[str, str]]:
+        """
+        ``GET https://SDK_BASE_URL/serving/serving_workflows/{workflow_key}/results``
+
+        Gets the workflow output result.
+
+        :returns: WorkflowOutputData
+
+        :rtype: string
+        """
+
+        endpoint = f"serving/serving_workflows/{self.wf_key}/results"
+        return self._make_request(endpoint, "GET")
--- a/WorkflowExecAgent/tools/custom_prompt.py
+++ b/WorkflowExecAgent/tools/custom_prompt.py
@@ -0,0 +1,9 @@
+# Copyright (C) 2024 Intel Corporation
+# SPDX-License-Identifier: Apache-2.0
+
+REACT_SYS_MESSAGE = """\
+You are a helpful assistant. You are to start the workflow using the tool provided.
+After the workflow is completed, you will use the output data to answer the user's original question in a one short sentence.
+
+Now begin!
+"""
--- a/WorkflowExecAgent/tools/sdk.py
+++ b/WorkflowExecAgent/tools/sdk.py
@@ -0,0 +1,19 @@
+# Copyright (C) 2024 Intel Corporation
+# SPDX-License-Identifier: Apache-2.0
+
+import os
+
+from tools.components.workflow import Workflow
+from tools.utils.handle_requests import RequestHandler
+
+
+class EasyDataSDK:
+    def __init__(self):
+        self.request_handler = RequestHandler(os.environ["SDK_BASE_URL"], os.environ["SERVING_TOKEN"])
+
+    def create_workflow(self, workflow_id=None, workflow_key=None):
+        return Workflow(
+            self.request_handler,
+            workflow_id=workflow_id,
+            workflow_key=workflow_key,
+        )
--- a/WorkflowExecAgent/tools/tools.py
+++ b/WorkflowExecAgent/tools/tools.py
@@ -0,0 +1,42 @@
+# Copyright (C) 2024 Intel Corporation
+# SPDX-License-Identifier: Apache-2.0
+
+import time
+
+from tools.sdk import EasyDataSDK
+
+
+def workflow_executor(params, workflow_id: int) -> dict:
+    sdk = EasyDataSDK()
+    workflow = sdk.create_workflow(workflow_id)
+
+    params = {key: str(val) for key, val in params.items()}
+    start_workflow = workflow.start(params)
+    print(start_workflow)
+
+    def check_workflow():
+        workflow_status = workflow.get_status()["workflow_status"]
+        if workflow_status == "finished":
+            message = "Workflow finished."
+        elif workflow_status == "initializing" or workflow_status == "running":
+            message = "Workflow execution is still in progress."
+        else:
+            message = "Workflow has failed."
+
+        return workflow_status, message
+
+    MAX_RETRY = 50
+    num_retry = 0
+    while num_retry < MAX_RETRY:
+        workflow_status, message = check_workflow()
+        print(message)
+        if workflow_status == "failed" or workflow_status == "finished":
+            break
+        else:
+            time.sleep(100)  # interval between each status checking retry
+            num_retry += 1
+
+    if workflow_status == "finished":
+        return workflow.result()
+    else:
+        return message
--- a/WorkflowExecAgent/tools/tools.yaml
+++ b/WorkflowExecAgent/tools/tools.yaml
@@ -0,0 +1,14 @@
+# Copyright (C) 2024 Intel Corporation
+# SPDX-License-Identifier: Apache-2.0
+
+workflow_executor:
+  description: "Starts a workflow with the given workflow id and params. Gets the output result of the workflow."
+  callable_api: tools.py:workflow_executor
+  args_schema:
+    workflow_id:
+      type: int
+      description: Workflow id
+    params:
+      type: dict
+      description: Workflow parameters.
+  return_output: workflow_data
--- a/WorkflowExecAgent/tools/utils/handle_requests.py
+++ b/WorkflowExecAgent/tools/utils/handle_requests.py
@@ -0,0 +1,76 @@
+# Copyright (C) 2024 Intel Corporation
+# SPDX-License-Identifier: Apache-2.0
+
+from functools import wraps
+
+import requests
+
+
+class RequestHandler:
+    """Class for handling requests.
+
+    Attributes:
+        base_url (string): The url of the API.
+        api_key (string): Secret token.
+    """
+
+    def __init__(self, base_url: str, api_key: str):
+        self.base_url = base_url
+        self.api_key = api_key
+
+    def _make_request(self, endpoint, method="GET", data=None, stream=False):
+        url = f"{self.base_url}{endpoint}"
+        headers = {"Content-Type": "application/json", "Authorization": f"Bearer {self.api_key}"}
+
+        error = ""
+
+        if method == "GET":
+            response = requests.get(url, headers=headers)
+        elif method == "POST":
+            response = requests.post(url, data, headers=headers, stream=stream)
+        elif method == "PUT":
+            response = requests.put(url, data, headers=headers)
+        elif method == "DELETE":
+            response = requests.delete(url, headers=headers)
+        else:
+            raise ValueError(f"error: Invalid HTTP method {method}")
+
+        @self._handle_request
+        def check_status(response):
+            response.raise_for_status()
+
+        error = check_status(response)
+
+        if error:
+            return error
+
+        else:
+            try:
+                response.json()
+                return response.json()
+            except:
+                return response
+
+    def _handle_request(self, func):
+        @wraps(func)
+        def decorated(response=None, *args, **kwargs):
+            if response is not None:
+                try:
+                    return func(response, *args, **kwargs)
+
+                except requests.exceptions.HTTPError as errh:
+                    error = {"error": f"{response.status_code} {response.reason} HTTP Error {errh}"}
+                except requests.exceptions.ConnectionError as errc:
+                    error = {"error": f"{response.status_code} {response.reason} Connection Error {errc}"}
+                except requests.exceptions.Timeout as errt:
+                    error = {"error": f"{response.status_code} {response.reason} Timeout Error {errt}"}
+                except requests.exceptions.ChunkedEncodingError as errck:
+                    error = {"error": f"Invalid chunk encoding: {str(errck)}"}
+                except requests.exceptions.RequestException as err:
+                    error = {"error": f"{response.status_code} {response.reason} {err}"}
+                except Exception as err:
+                    error = err
+
+                return error
+
+        return decorated