KasarLabs · ijusttookadnatest · Jun 10, 2025 · Jun 10, 2025 · Jun 10, 2025 · Jun 10, 2025
diff --git a/.env.example b/.env.example
@@ -0,0 +1,3 @@
+POSTGRES_USER="cairocoder"
+POSTGRES_PASSWORD="cairocoder"
+POSTGRES_DB="cairocoder"
diff --git a/.github/workflows/build.yml b/.github/workflows/build.yml
@@ -69,7 +69,7 @@ jobs:
         with:
           context: .
           push: true
-          file: ./ingest.dockerfile
+          file: ./ingester.dockerfile
           tags: ${{ steps.meta-ingester.outputs.tags }}
           labels: ${{ steps.meta-ingester.outputs.labels }}
           cache-from: type=gha

diff --git a/.github/workflows/generate-embeddings.yml b/.github/workflows/generate-embeddings.yml
@@ -87,7 +87,7 @@ jobs:
           PORT = 3001
           SIMILARITY_MEASURE = "cosine"
 
-          [HOSTED_MODE]
+          [PROVIDERS]
           DEFAULT_CHAT_PROVIDER = "${{ secrets.DEFAULT_CHAT_PROVIDER }}"
           DEFAULT_CHAT_MODEL = "${{ secrets.DEFAULT_CHAT_MODEL }}"
           DEFAULT_FAST_CHAT_PROVIDER = "${{ secrets.DEFAULT_FAST_CHAT_PROVIDER }}"

diff --git a/README.md b/README.md
@@ -14,9 +14,7 @@
   - [Project Structure](#project-structure)
   - [RAG Pipeline](#rag-pipeline)
   - [Ingestion System](#ingestion-system)
-  - [Database](#database)
 - [Development](#development)
-- [Upcoming Features](#upcoming-features)
 - [Contribution](#contribution)
 
 ## Credits
@@ -34,7 +32,6 @@ Cairo Coder is an intelligent code generation service that makes writing Cairo s
 - **OpenAI Compatible API**: Interface compatible with the OpenAI API format for easy integration
 - **Multiple LLM Support**: Works with OpenAI, Anthropic, and Google models
 - **Source-Informed Generation**: Code is generated based on Cairo documentation, ensuring correctness
-- **Streaming Response**: Support for response streaming for a responsive experience
 
 
 ## Installation
@@ -64,12 +61,12 @@ There are mainly 2 ways of installing Cairo Coder - With Docker, Without Docker.
 5. Inside the packages/agents package, copy the `sample.config.toml` file to a `config.toml`. For development setups, you need only fill in the following fields:
 
    - `OPENAI`: Your OpenAI API key. **You only need to fill this if you wish to use OpenAI's models**.
-   - `ANTHROPIC`: Your Anthropic API key. **You only need to fill this if you wish to use Anthropic models**.
+   - `GEMINI`: Your Gemini API key. **You only need to fill this if you wish to use Gemini models**.
    - `SIMILARITY_MEASURE`: The similarity measure to use (This is filled by default; you can leave it as is if you are unsure about it.)
-   - Models: The `[HOSTED_MODE]` table defines the underlying LLM model used. We recommend using:
+   - Models: The `[PROVIDERS]` table defines the underlying LLM model used. We recommend using:
 
    ```toml
-      [HOSTED_MODE]
+      [PROVIDERS]
       DEFAULT_CHAT_PROVIDER = "gemini"
       DEFAULT_CHAT_MODEL = "Gemini Flash 2.5"
       DEFAULT_FAST_CHAT_PROVIDER = "gemini"
@@ -83,19 +80,15 @@ There are mainly 2 ways of installing Cairo Coder - With Docker, Without Docker.
    Cairo Coder uses PostgreSQL with pgvector for storing and retrieving vector embeddings. You need to configure both the database initialization and the application connection settings:
 
    **a. Database Container Initialization** (`.env` file):
-
    Create a `.env` file in the root directory with the following PostgreSQL configuration:
 
    ```
    POSTGRES_USER="YOUR_POSTGRES_USER"
    POSTGRES_PASSWORD="YOUR_POSTGRES_PASSWORD"
-   POSTGRES_ROOT_DB="YOUR_POSTGRES_ROOT_DB"
-   POSTGRES_HOST="localhost"
-   POSTGRES_PORT="5432"
+   POSTGRES_DB="YOUR_POSTGRES_DB"
    ```
 
    This file is used by Docker to initialize the PostgreSQL container when it first starts.
-   The `POSTGRES_HOST` is set to "localhost" because this is from the database's own perspective.
 
    **b. Application Connection Settings** (`config.toml` file):
 
@@ -105,16 +98,15 @@ There are mainly 2 ways of installing Cairo Coder - With Docker, Without Docker.
     [VECTOR_DB]
     POSTGRES_USER="YOUR_POSTGRES_USER"
     POSTGRES_PASSWORD="YOUR_POSTGRES_PASSWORD"
-    POSTGRES_ROOT_DB="YOUR_POSTGRES_ROOT_DB"
+    POSTGRES_DB="YOUR_POSTGRES_DB"
     POSTGRES_HOST="postgres"
     POSTGRES_PORT="5432"
    ```
 
    This configuration is used by the backend and ingester services to connect to the database.
-   Note that `POSTGRES_HOST` is set to "postgres", which is the service name in docker-compose.yml.
+   Note that `POSTGRES_HOST` is set to ```"postgres"``` and `POSTGRES_PORT` to ```"5432"```, which are the container's name and port in docker-compose.yml.
 
-   **Important:** Make sure to use the same password in both files. The first file initializes the
-   database, while the second is used by your application to connect to it.
+   **Important:** Make sure to use the same password, username and db's name in both files. The first file initializes the database, while the second is used by your application to connect to it.
 
 
 7. **Configure LangSmith (Optional)**
@@ -131,24 +123,25 @@ There are mainly 2 ways of installing Cairo Coder - With Docker, Without Docker.
    LANGSMITH_API_KEY="<your-api-key>"
    LANGCHAIN_PROJECT="<your-project-name>"
    ```
+   - Add the `.env` in an env_file section in the backend service of the docker-compose.yml 
 
    With this configuration, all LLM calls and chain executions will be logged to your LangSmith project, allowing you to debug, analyze, and improve the system's performance.
 
 
 9. Run the application using one of the following methods:
 
    ```bash
-   docker-compose up --build
+   docker compose up postgres backend
    ```
 
-8. The API will be available at http://localhost:3000/chat/completions.
+8. The API will be available at http://localhost:3001/v1/chat/completions
 
 ## Running the Ingester
 
 After you have the main application running, you might need to run the ingester to process and embed documentation from various sources. The ingester is configured as a separate profile in the docker-compose file and can be executed as follows:
 
    ```bash
-   docker-compose --profile ingester up ingester
+   docker compose up ingester
    ```
 
 Once the ingester completes its task, the vector database will be populated with embeddings from all the supported documentation sources, making them available for RAG-based code generation requests to the API.
@@ -160,46 +153,71 @@ Cairo Coder provides a simple REST API compatible with the OpenAI format for eas
 ### Endpoint
 
 ```
-POST /chat/completions
+POST /v1/chat/completions
 ```
 
 ### Request Format
 
-```json
-{
-  "model": "gemini-2.0-flash",
-  "messages": [
-    {
-      "role": "system",
-      "content": "You are a Cairo programming expert."
-    },
-    {
-      "role": "user",
-      "content": "Write a Cairo contract that implements a simple ERC-20 token."
-    }
-  ],
-  "temperature": 0.7,
-}
+Example of a simple request:
+
+```bash
+curl -X POST http://localhost:3001/v1/chat/completions \
+  -H "Content-Type: application/json" \
+  -d '{
+    "model": "gemini-2.5-flash",
+    "messages": [
+      {
+        "role": "user",
+        "content": "How do I implement storage in Cairo?"
+      }
+    ]
+  }'
 ```
 
+The API accepts all standard OpenAI Chat Completions parameters.
+
+**Supported Parameters:**
+- `model`: Model identifier (string)
+- `messages`: Array of message objects with `role` and `content`
+- `temperature`: Controls randomness (0-2, default: 0.7)
+- `top_p`: Nucleus sampling parameter (0-1, default: 1)
+- `n`: Number of completions (default: 1)
+- `stream`: Enable streaming responses (boolean, default: false)
+- `max_tokens`: Maximum tokens in response
+- `stop`: Stop sequences (string or array)
+- `presence_penalty`: Penalty for token presence (-2 to 2)
+- `frequency_penalty`: Penalty for token frequency (-2 to 2)
+- `logit_bias`: Token bias adjustments
+- `user`: User identifier
+- `response_format`: Response format specification
+
+
 ### Response Format
 
+#### Standard Mode Response
+
 ```json
 {
-  "id": "gen-123456",
+  "id": "chatcmpl-123456",
   "object": "chat.completion",
   "created": 1717273561,
-  "model": "gemini-2.0-flash",
+  "model": "gemini-2.5-flash",
   "choices": [
     {
       "index": 0,
       "message": {
         "role": "assistant",
-        "content": "#[starknet::contract]\nmod ERC20 {\n    // Contract code here...\n}"
+        "content": "#[starknet::contract]\nmod ERC20 {\n    use starknet::storage::{StoragePointerReadAccess, StoragePointerWriteAccess};\n    \n    #[storage]\n    struct Storage {\n        name: felt252,\n        symbol: felt252,\n        total_supply: u256,\n        balances: Map<ContractAddress, u256>,\n    }\n    // ... contract implementation\n}"
       },
+      "logprobs": null,
       "finish_reason": "stop"
     }
-  ]
+  ],
+  "usage": {
+    "prompt_tokens": 45,
+    "completion_tokens": 120,
+    "total_tokens": 165
+  }
 }
 ```
 
@@ -245,13 +263,6 @@ Currently supported documentation sources include:
 - Cairo Foundry documentation
 - Cairo By Examples
 
-### Database
-
-Cairo Coder uses MongoDB Atlas with vector search capabilities for similarity search:
-
-- **Vector Database**: Stores document embeddings for efficient similarity search
-- **Vector Search**: Uses cosine similarity to find relevant Cairo documentation
-
 ## Development
 
 For development, you can use the following commands:

diff --git a/docker-compose.yml b/docker-compose.yml
@@ -1,14 +1,12 @@
-version: '3.8'
-
 services:
   postgres:
     image: pgvector/pgvector:pg17
-    container_name: "postgresql"
+    container_name: 'postgres'
     shm_size: 1g
     env_file:
       - .env
     ports:
-      - 5432:5432
+      - 5455:5432
     volumes:
       - ./data:/var/lib/postgresql/data
     restart: unless-stopped
@@ -19,12 +17,9 @@ services:
     build:
       context: .
       dockerfile: backend.dockerfile
+    container_name: 'cairo-coder-backend'
     ports:
       - 3001:3001
-    extra_hosts:
-      - host.docker.internal:host-gateway
-    env_file:
-      - packages/backend/.env
     depends_on:
       postgres:
         condition: service_started
@@ -34,10 +29,10 @@ services:
 
   ingester:
     platform: linux/amd64
+    container_name: 'cairo-coder-ingester'
     build:
       context: .
-      dockerfile: ingest.dockerfile
-    profiles: ["ingester"]
+      dockerfile: ingester.dockerfile
     depends_on:
       postgres:
         condition: service_started

diff --git a/ingest.dockerfile → ingester.dockerfile b/ingest.dockerfile → ingester.dockerfile
diff --git a/packages/agents/.env.test b/packages/agents/.env.test
diff --git a/packages/agents/sample.config.toml b/packages/agents/sample.config.toml
@@ -5,21 +5,18 @@ ANTHROPIC = ""
 DEEPSEEK = ""
 GEMINI = ""
 
-[API_ENDPOINTS]
-OLLAMA = ""
-
 [VECTOR_DB]
 POSTGRES_USER="cairocoder"
 POSTGRES_PASSWORD="cairocoder"
-POSTGRES_ROOT_DB="cairocoder"
-POSTGRES_HOST="localhost"
+POSTGRES_DB="cairocoder"
+POSTGRES_HOST="postgres"
 POSTGRES_PORT="5432"
 
 [GENERAL]
 PORT = 3_001
 SIMILARITY_MEASURE = "cosine"
 
-[HOSTED_MODE]
+[PROVIDERS]
 DEFAULT_CHAT_PROVIDER = "gemini"
 DEFAULT_CHAT_MODEL = "Gemini Flash 2.5"
 DEFAULT_FAST_CHAT_PROVIDER = "gemini"

diff --git a/packages/agents/src/config/settings.ts b/packages/agents/src/config/settings.ts
@@ -26,7 +26,7 @@ const loadConfig = () => {
   ) as any as Config;
 };
 
-export const getHostedModeConfig = () => loadConfig().HOSTED_MODE;
+export const getHostedModeConfig = () => loadConfig().PROVIDERS;
 
 export const getPort = () => loadConfig().GENERAL.PORT;
 
@@ -49,7 +49,7 @@ export const getVectorDbConfig = () => {
   return {
     POSTGRES_USER: config.VECTOR_DB.POSTGRES_USER || '',
     POSTGRES_PASSWORD: config.VECTOR_DB.POSTGRES_PASSWORD || '',
-    POSTGRES_ROOT_DB: config.VECTOR_DB.POSTGRES_ROOT_DB || '',
+    POSTGRES_DB: config.VECTOR_DB.POSTGRES_DB || '',
     POSTGRES_HOST: config.VECTOR_DB.POSTGRES_HOST || '',
     POSTGRES_PORT: config.VECTOR_DB.POSTGRES_PORT || '',
   } as VectorStoreConfig;

diff --git a/packages/agents/src/db/postgresVectorStore.ts b/packages/agents/src/db/postgresVectorStore.ts
@@ -108,7 +108,7 @@ export class VectorStore {
       const pool = new Pool({
         user: config.POSTGRES_USER, 
         host: config.POSTGRES_HOST,
-        database: config.POSTGRES_ROOT_DB,
+        database: config.POSTGRES_DB,
         password: config.POSTGRES_PASSWORD,
         port: parseInt(config.POSTGRES_PORT),
         max: 10,

diff --git a/packages/agents/src/types/index.ts b/packages/agents/src/types/index.ts
@@ -13,7 +13,7 @@ export interface LLMConfig {
 export interface VectorStoreConfig {
   POSTGRES_USER: string;
   POSTGRES_PASSWORD: string;
-  POSTGRES_ROOT_DB: string;
+  POSTGRES_DB: string;
   POSTGRES_HOST: string;
   POSTGRES_PORT: string;
 }
@@ -31,7 +31,7 @@ export interface Config {
     DEEPSEEK: string;
     GEMINI: string;
   };
-  HOSTED_MODE?: {
+  PROVIDERS?: {
     DEFAULT_CHAT_PROVIDER: string;
     DEFAULT_CHAT_MODEL: string;
     DEFAULT_FAST_CHAT_PROVIDER: string;

diff --git a/packages/backend/src/config/provider/index.ts b/packages/backend/src/config/provider/index.ts
@@ -55,13 +55,13 @@ export const getAvailableEmbeddingModelProviders = async () => {
   }
 
   const hostedModeConfig = getHostedModeConfig();
-  const hosted_model =
+  const PROVIDERSl =
     models[hostedModeConfig.DEFAULT_EMBEDDING_PROVIDER][
       hostedModeConfig.DEFAULT_EMBEDDING_MODEL
     ];
   return {
     [hostedModeConfig.DEFAULT_EMBEDDING_PROVIDER]: {
-      [hostedModeConfig.DEFAULT_EMBEDDING_MODEL]: hosted_model,
+      [hostedModeConfig.DEFAULT_EMBEDDING_MODEL]: PROVIDERSl,
     },
   };
 };