Data

Tools

Indexes

Applications

Experiments

Awesome

An open API service indexing awesome lists of open source software.

https://github.com/code-monad/prompt-wrapper


https://github.com/code-monad/prompt-wrapper

Last synced: 10 months ago
JSON representation

Awesome Lists containing this project

README 

          
# Prompt Wrapper Service



A simple service that provides wise sayings from an LLM (OpenRouter) with rate limiting and caching.



## Features



- RESTful HTTP endpoints for managing sayings and presets

- Rate limiting for users

- Persistent storage with Sled embedded database

- Rich UI-ready preset configurations

- Random preset selection for each user session

- Integration with OpenRouter for LLM capabilities

- Test user available in debug builds for easy testing



## Setup



1. Clone the repository

2. Copy `.env.example` to `.env` and update with your settings

3. Get an API key from [OpenRouter](https://openrouter.ai)

4. Update the `.env` file with your OpenRouter API key

5. Customize prompt presets in `presets.yaml` file



## Running the service



```bash

cargo run

```



The service will be available at http://localhost:3000



## API Endpoints



### Sayings Resource



#### GET /sayings



Returns a list of sayings for the specified user.



**Query Parameters:**

- `user_id` (optional): Identifier for the user. If not provided, a default value is used.

- `limit` (optional): Maximum number of sayings to return. Default is 10.



**Response:**

```json

[

  {

    "id": "uuid1",

    "content": "Saying content 1",

    "created_at": "2023-01-01T00:00:00Z",

    "source": "llm"

  },

  {

    "id": "uuid2",

    "content": "Saying content 2",

    "created_at": "2023-01-01T00:00:00Z",

    "source": "llm"

  }

]

```



#### GET /sayings/latest



Returns the latest saying for the specified user.



**Query Parameters:**

- `user_id` (optional): Identifier for the user. If not provided, a default value is used.



**Response:**

```json

{

  "id": "uuid",

  "content": "The saying content",

  "created_at": "2023-01-01T00:00:00Z",

  "source": "llm"

}

```



#### POST /sayings



Creates a new saying using the OpenRouter LLM API and returns it.



**Query Parameters:**

- `user_id` (optional): Identifier for the user. If not provided, a default value is used.



**Request Body:**

```json

{

  "prompt": "Optional prompt to guide the LLM",

  "preset_id": "Optional preset ID to use a specific preset"

}

```



If neither `prompt` nor `preset_id` is provided, the service will use the preset that was randomly selected for the user.



**Response:**

```json

{

  "id": "uuid",

  "content": "The saying content",

  "created_at": "2023-01-01T00:00:00Z",

  "source": "llm"

}

```



### User Status Resource



#### GET /users/{user_id}/status



Returns the user's rate limit status, their last retrieved saying, and their currently selected preset.



**Response:**

```json

{

  "user_id": "user123",

  "can_query": true,

  "remaining_requests": 5,

  "reset_at": "2023-01-01T01:00:00Z",

  "last_saying": {

    "id": "uuid",

    "content": "The last saying content",

    "created_at": "2023-01-01T00:00:00Z",

    "source": "llm"

  },

  "selected_preset": {

    "id": "oracle",

    "name": "Ape Oracle",

    "description": "Ancient wisdom for modern questions",

    "tags": ["oracle", "wisdom", "mystical"],

    "button_text": "Reveal My Answer",

    "loading_text": "Consulting the oracle...",

    "instruction_text": "Focus on your question, press the button, and receive wisdom from the Ape Oracle..."

  }

}

```



### Presets Resource



#### GET /presets



Returns all available presets.



**Response:**

```json

[

  {

    "id": "oracle",

    "name": "Ape Oracle",

    "description": "Ancient wisdom for modern questions",

    "tags": ["oracle", "wisdom", "mystical"],

    "button_text": "Reveal My Answer",

    "loading_text": "Consulting the oracle...",

    "instruction_text": "Focus on your question, press the button, and receive wisdom from the Ape Oracle..."

  },

  {

    "id": "fortune",

    "name": "Fortune Teller",

    "description": "Glimpse into your future",

    "tags": ["fortune", "future", "prediction"],

    "button_text": "Read My Fortune",

    "loading_text": "Gazing into the crystal ball...",

    "instruction_text": "Think of what you wish to know about your future, then press the button..."

  }

]

```



#### GET /presets/{preset_id}



Returns a specific preset by ID.



**Response:**

```json

{

  "id": "oracle",

  "name": "Ape Oracle",

  "description": "Ancient wisdom for modern questions",

  "tags": ["oracle", "wisdom", "mystical"],

  "button_text": "Reveal My Answer",

  "loading_text": "Consulting the oracle...",

  "instruction_text": "Focus on your question, press the button, and receive wisdom from the Ape Oracle..."

}

```



## Presets Configuration



Presets are defined in a YAML file specified by the `PRESETS_FILE_PATH` environment variable. Each preset contains:



- `id`: Unique identifier for the preset

- `name`: Display name for the preset

- `description`: Short description of the preset's purpose

- `tags`: List of category tags

- `button_text`: Text to display on action buttons

- `loading_text`: Text to display during loading/processing

- `instruction_text`: Guidance text for users

- `system_prompt`: The system prompt to set the context for the LLM

- `user_prompts`: List of possible user prompts that will be randomly selected



Example preset configuration:



```yaml

- id: oracle

  name: Ape Oracle

  description: Ancient wisdom for modern questions

  tags:

    - oracle

    - wisdom

    - mystical

  button_text: Reveal My Answer

  loading_text: Consulting the oracle...

  instruction_text: Focus on your question, press the button, and receive wisdom from the Ape Oracle...

  system_prompt: >

    You are the Ape Oracle, a mystical entity that provides profound, 

    wise, and sometimes cryptic answers to unspoken questions.

  user_prompts:

    - "Will I find success?"

    - "What should I do next?"

    - "Is this the right path?"

```



## Configuration



All configuration is done through environment variables or the `.env` file:



- `SERVER_HOST`: Host to bind the server to

- `SERVER_PORT`: Port to bind the server to

- `OPENROUTER_API_KEY`: Your OpenRouter API key

- `OPENROUTER_MODEL`: The model to use (default: mistralai/mistral-7b-instruct)

- `RATE_LIMIT_MAX_REQUESTS`: Maximum number of requests per window

- `RATE_LIMIT_WINDOW_SECONDS`: Window size in seconds for rate limiting

- `STORAGE_TYPE`: Type of storage to use (memory, sled, redis, sqlite)

- `STORAGE_CONNECTION_STRING`: Connection string for the storage

- `PRESETS_FILE_PATH`: Path to the presets YAML file



## Development Features



### Test User



In debug builds (when compiling without the `--release` flag), a test user is automatically initialized with:



- User ID: `test_user` 

- Empty initial state (no pre-populated sayings)

- Fresh rate limit quota (according to configured settings)



The test user follows a completely dynamic workflow identical to regular users:

- All sayings are generated on-demand via the LLM or cache

- Rate limiting follows standard rules

- Presets are dynamically selected for each user session

- No hardcoded or static responses are used



You can use this test user for development and testing purposes by including `user_id=test_user` in your requests:



```

GET /sayings?user_id=test_user

```



**Note:** This test user ID is blocked in release/production builds to prevent misuse in production environments.

To test in release mode, use a different user ID.



## License



MIT