Upstream E-RAG's prompt template microservice.

.github/workflows/docker/compose/prompt_template-compose.yaml

@@ -0,0 +1,9 @@
+# Copyright (C) 2025 Intel Corporation
+# SPDX-License-Identifier: Apache-2.0
+
+# this file should be run in the root of the repo
+services:
+  prompt-template:
+    build:
+      dockerfile: comps/prompt_template/src/Dockerfile
+    image: ${REGISTRY:-opea}/prompt-template:${TAG:-latest}

comps/cores/mega/constants.py

@@ -39,6 +39,7 @@ class ServiceType(Enum):
     TEXT2KG = 22
     STRUCT2GRAPH = 23
     LANGUAGE_DETECTION = 24
+    PROMPT_TEMPLATE = 25
 
 
 class MegaServiceEndpoint(Enum):

comps/prompt_template/deployment/docker_compose/compose.yaml

@@ -0,0 +1,19 @@
+# Copyright (C) 2025 Intel Corporation
+# SPDX-License-Identifier: Apache-2.0
+
+services:
+  prompt-template:
+    image: ${REGISTRY:-opea}/prompt-template:${TAG:-latest}
+    container_name: prompt-template
+    ports:
+      - "7900:7900"
+    environment:
+      - no_proxy=${no_proxy}
+      - https_proxy=${https_proxy}
+      - http_proxy=${http_proxy}
+    ipc: host
+    restart: always
+
+networks:
+  default:
+    driver: bridge

comps/prompt_template/src/Dockerfile

@@ -0,0 +1,34 @@
+# Copyright (C) 2024 Intel Corporation
+# SPDX-License-Identifier: Apache-2.0
+
+# FROM python:3.11-slim
+FROM python:3.11-slim
+
+# Set environment variables
+ENV LANG=en_US.UTF-8
+
+RUN apt-get update -y && \
+    apt-get install build-essential -y && \
+    apt-get install -y --no-install-recommends --fix-missing \
+    vim && \
+    apt-get clean && rm -rf /var/lib/apt/lists/*
+
+COPY comps /home/comps
+
+RUN useradd -m -s /bin/bash user && \
+    mkdir -p /home/user && \
+    chown -R user /home/user/
+
+ARG uvpip='uv pip install --system --no-cache-dir'
+RUN pip install --no-cache-dir --upgrade pip setuptools uv && \
+    $uvpip -r /home/comps/prompt_template/src/requirements.txt
+
+ENV PYTHONPATH=$PYTHONPATH:/home
+
+USER user
+
+WORKDIR /home/comps/prompt_template/src
+
+ENTRYPOINT ["python", "opea_prompt_template_microservice.py"]
+
+

comps/prompt_template/src/README.md

@@ -0,0 +1,199 @@
+# Prompt Template microservice
+
+The Prompt Template microservice dynamically generates system and user prompts based on structured inputs and document context. It supports usage in LLM pipelines to customize prompt formatting with reranked documents, conversation history, and user queries.
+
+## Getting started
+
+### 🚀1. Start Prompt Template Microservice with Python (Option 1)
+
+To start the Prompt Template microservice, you need to install Python packages first.
+
+#### 1.1. Install Requirements
+
+```bash
+pip install -r requirements.txt
+```
+
+#### 1.2. Start Microservice
+
+```bash
+python opea_prompt_template_microservice.py
+```
+
+### 🚀2. Start Prompt Template Microservice with Docker (Option 2)
+
+#### 2.1. Build the Docker Image:
+
+Use the below docker build command to create the image:
+
+```bash
+cd ../../../
+docker build -t opea/prompt-template:latest -f comps/prompt_template/src/Dockerfile .
+```
+
+Please note that the building process may take a while to complete.
+
+#### 2.2. Run the Docker Container:
+
+```bash
+docker run -d --name="prompt-template-microservice" \
+  -p 7900:7900 \
+  --net=host \
+  --ipc=host \
+  opea/prompt-template:latest
+```
+
+### 3. Verify the Prompt Template Microservice
+
+#### 3.1. Check Status
+
+```bash
+curl http://localhost:7900/v1/health_check \
+  -X GET \
+  -H 'Content-Type: application/json'
+```
+
+#### 3.2. Sending a Request
+
+##### 3.2.1 Default Template Generation
+
+Generates the prompt using the default template:
+
+**Example Input**
+
+```bash
+curl -X POST -H "Content-Type: application/json" -d @- http://localhost:7900/v1/prompt_template <<JSON_DATA
+{
+  "data": {
+    "user_prompt": "What is Deep Learning?",
+    "reranked_docs": [{ "text": "Deep Learning is a subfield of machine learning..." }]
+  },
+  "conversation_history": [
+    { "question": "Hello", "answer": "Hello as well" },
+    { "question": "How are you?", "answer": "I am good, thank you!" }
+  ],
+  "system_prompt_template": "",
+  "user_prompt_template": ""
+}
+JSON_DATA
+```
+
+**Example Output**
+
+A chat_template starting with the default assistant description.
+
+```json
+{
+  "id": "4e799abdf5f09433adc276b511a8b0ae",
+  "model": null,
+  "query": "What is Deep Learning?",
+  "max_tokens": 1024,
+  "max_new_tokens": 1024,
+  "top_k": 10,
+  "top_p": 0.95,
+  "typical_p": 0.95,
+  "temperature": 0.01,
+  "frequency_penalty": 0.0,
+  "presence_penalty": 0.0,
+  "repetition_penalty": 1.03,
+  "stream": true,
+  "language": "auto",
+  "input_guardrail_params": null,
+  "output_guardrail_params": null,
+  "chat_template": "### You are a helpful, respectful, and honest assistant to help the user with questions. Please refer to the search results obtained from the local knowledge base. Refer also to the conversation history if you think it is relevant to the current question. Ignore all information that you think is not relevant to the question. If you don'\''t know the answer to a question, please don'\''t share false information. \n ### Search results: [File: Unknown Source]\nDeep Learning is...\n### Conversation history: User: Hello\nAssistant: Hello as well\nUser: How are you?\nAssistant: I am good, thank you!\nUser: Who are you?\nAssistant: I am a robot\n### Question: What is Deep Learning? \n\n### Answer:",
+  "documents": []
+}
+```
+
+##### 3.2.2 Custom Prompt Template
+
+You can provide custom system and user prompt templates:
+
+**Example Input**
+
+```bash
+curl -X POST -H "Content-Type: application/json" -d @- http://localhost:7900/v1/prompt_template <<JSON_DATA
+{
+  "data": {
+    "initial_query": "What is Deep Learning?",
+    "reranked_docs": [{ "text": "Deep Learning is..." }]
+  },
+  "system_prompt_template": "### Please refer to the search results obtained from the local knowledge base. But be careful to not incorporate information that you think is not relevant to the question. If you don't know the answer to a question, please don't share false information. ### Search results: {reranked_docs}",
+  "user_prompt_template": "### Question: {initial_query} \\n### Answer:"
+}
+JSON_DATA
+```
+
+**Example Output**
+
+Custom instructions about using search results in the chat_template.
+
+```json
+{
+  "id": "b1f1cec396954d5dc1b942f5959d556d",
+  "model": null,
+  "query": "What is Deep Learning?",
+  "max_tokens": 1024,
+  "max_new_tokens": 1024,
+  "top_k": 10,
+  "top_p": 0.95,
+  "typical_p": 0.95,
+  "temperature": 0.01,
+  "frequency_penalty": 0.0,
+  "presence_penalty": 0.0,
+  "repetition_penalty": 1.03,
+  "stream": true,
+  "language": "auto",
+  "input_guardrail_params": null,
+  "output_guardrail_params": null,
+  "chat_template": "### Please refer to the search results obtained from the local knowledge base. But be careful to not incorporate information that you think is not relevant to the question. If you don'\''t know the answer to a question, please don'\''t share false information. ### Search results: [File: Unknown Source]\nDeep Learning is...\n### Question: What is Deep Learning? \n### Answer:",
+  "documents": []
+}
+```
+
+##### 3.2.3 Translation Scenario
+
+Using a translation-related prompt template:
+
+**Example Input**
+
+```bash
+curl -X POST -H "Content-Type: application/json" -d @- http://localhost:7900/v1/prompt_template <<JSON_DATA
+{
+  "data": {
+    "initial_query": "What is Deep Learning?",
+    "source_lang": "chinese",
+    "target_lang": "english"
+  },
+  "system_prompt_template": "### You are a helpful, respectful, and honest assistant to help the user with translations. Translate this from {source_lang} to {target_lang}.",
+  "user_prompt_template": "### Question: {initial_query} \\n### Answer:"
+}
+JSON_DATA
+```
+
+**Example Output**
+
+A translation instruction like: Translate this from chinese to english.
+
+```json
+{
+  "id": "4f5e0024c2330a7be065b370d02e061f",
+  "model": null,
+  "query": "什么是深度学习？",
+  "max_tokens": 1024,
+  "max_new_tokens": 1024,
+  "top_k": 10,
+  "top_p": 0.95,
+  "typical_p": 0.95,
+  "temperature": 0.01,
+  "frequency_penalty": 0.0,
+  "presence_penalty": 0.0,
+  "repetition_penalty": 1.03,
+  "stream": true,
+  "language": "auto",
+  "input_guardrail_params": null,
+  "output_guardrail_params": null,
+  "chat_template": "### You are a helpful, respectful, and honest assistant to help the user with translations. Translate this from chinese to english.\n### Question: 什么是深度学习？ \n### Answer:",
+  "documents": []
+}
+```

comps/prompt_template/src/integrations/__init__.py

@@ -0,0 +1,2 @@
+# Copyright (C) 2024 Intel Corporation
+# SPDX-License-Identifier: Apache-2.0