feat: mcp added to http transport

Author: Pratham-Mishra04

.gitignore

@@ -3,5 +3,5 @@
 .DS_Store
 *_creds*
 **/venv/
-**/__pycache__/
-**/__pycache__/**
+**/__pycache__/**
+private.*

README.md

@@ -18,14 +18,16 @@ Bifrost is an open-source middleware that serves as a unified gateway to various
 
    ```json
    {
-     "openai": {
-       "keys": [
-         {
-           "value": "env.OPENAI_API_KEY",
-           "models": ["gpt-4o-mini"],
-           "weight": 1.0
-         }
-       ]
+     "providers": {
+       "openai": {
+         "keys": [
+           {
+             "value": "env.OPENAI_API_KEY",
+             "models": ["gpt-4o-mini"],
+             "weight": 1.0
+           }
+         ]
+       }
      }
    }
    ```
@@ -116,6 +118,7 @@ For additional configurations in HTTP server setup, please read [this](https://g
          {
            Value:  os.Getenv("OPENAI_API_KEY"),
            Models: []string{"gpt-4o-mini"},
+           Weight: 1.0,
          },
        }, nil
    }
@@ -205,12 +208,12 @@ With Bifrost, you can focus on building your AI-powered applications without wor
 
 - **Multi-Provider Support**: Integrate with OpenAI, Anthropic, Amazon Bedrock, Mistral, Ollama, and more through a single API
 - **Fallback Mechanisms**: Automatically retry failed requests with alternative models or providers
-- **Dynamic Key Management**: Rotate and manage API keys efficiently
+- **Dynamic Key Management**: Rotate and manage API keys efficiently with weighted distribution
 - **Connection Pooling**: Optimize network resources for better performance
 - **Concurrency Control**: Manage rate limits and parallel requests effectively
 - **Flexible Transports**: Multiple transports for easy integration into your infra
 - **Plugin First Architecture**: No callback hell, simple addition/creation of custom plugins
-- **MCP Integration**: Built-in Model Context Protocol (MCP) support for external tool integration
+- **MCP Integration**: Built-in Model Context Protocol (MCP) support for external tool integration and execution
 - **Custom Configuration**: Offers granular control over pool sizes, network retry settings, fallback providers, and network proxy configurations
 - **Built-in Observability**: Native Prometheus metrics out of the box, no wrappers, no sidecars, just drop it in and scrape
 

core/mcp.go

@@ -486,6 +486,8 @@ func (m *MCPManager) connectToMCPClient(config schemas.MCPClientConfig) error {
 		for toolName, tool := range tools {
 			client.ToolMap[toolName] = tool
 		}
+
+		m.logger.Info(fmt.Sprintf("%s Connected to MCP client: %s", MCPLogPrefix, config.Name))
 	} else {
 		return fmt.Errorf("client %s was removed during connection setup", config.Name)
 	}

docs/http-transport-api.md

@@ -4,9 +4,11 @@ This document provides comprehensive API documentation for the Bifrost HTTP tran
 
 ## Base URL
 
+```text
+  http://localhost:8080
 ```
-http://localhost:8080
-```
+
+> 🔧 **MCP (Model Context Protocol) Integration**: Bifrost HTTP transport includes built-in MCP support for external tool integration. When MCP is configured, tools are automatically discovered and added to model requests. For comprehensive MCP setup and usage, see the [**MCP Integration Guide**](mcp.md) and [**HTTP Transport MCP Configuration**](../transports/README.md#mcp-model-context-protocol-configuration).
 
 ## OpenAPI Specification
 
@@ -215,7 +217,86 @@ Creates a text completion from a prompt.
 }
 ```
 
-### 3. Metrics
+### 3. MCP Tool Execution
+
+**POST** `/v1/mcp/tool/execute`
+
+Executes MCP (Model Context Protocol) tools that have been configured in Bifrost. This endpoint is used to execute tool calls returned by AI models during conversations.
+
+> **Note**: This endpoint requires MCP to be configured in Bifrost. See [MCP Integration Guide](mcp.md) for setup instructions.
+
+#### Request Body
+
+```json
+{
+  "type": "function",
+  "id": "toolu_01Vmq4gaU6tSy7ZRKVC7U2fg",
+  "function": {
+    "name": "google_search",
+    "arguments": "{\"gl\":\"us\",\"hl\":\"en\",\"num\":5,\"q\":\"San Francisco news yesterday\",\"tbs\":\"qdr:d\"}"
+  }
+}
+```
+
+#### Response
+
+```json
+{
+  "role": "tool",
+  "content": "{\n  \"searchParameters\": {\n    \"q\": \"San Francisco news yesterday\",\n    \"gl\": \"us\",\n    \"hl\": \"en\",\n    \"type\": \"search\",\n    \"num\": 5,\n    \"tbs\": \"qdr:d\",\n    \"engine\": \"google\"\n  },\n  \"organic\": [\n    {\n      \"title\": \"San Francisco Chronicle · Giants' today\"\n    },\n    {\n      \"query\": \"s.f. chronicle e edition\"\n    }\n  ],\n  \"credits\": 1\n}",
+  "tool_call_id": "toolu_01Vmq4gaU6tSy7ZRKVC7U2fg"
+}
+```
+
+#### Multi-Turn Tool Workflow
+
+The typical workflow for using MCP tools involves:
+
+1. **Send chat completion request** → AI responds with `tool_calls`
+2. **Execute tools via `/v1/mcp/tool/execute`** → Get tool result messages
+3. **Add tool results to conversation** → Send back for final response
+
+```bash
+# Step 1: Chat completion (AI decides to use tools)
+curl -X POST http://localhost:8080/v1/chat/completions \
+  -H "Content-Type: application/json" \
+  -d '{
+    "provider": "openai",
+    "model": "gpt-4o-mini",
+    "messages": [
+      {"role": "user", "content": "Search for San Francisco news from yesterday"}
+    ]
+  }'
+
+# Step 2: Execute the tool call returned by AI
+curl -X POST http://localhost:8080/v1/mcp/tool/execute \
+  -H "Content-Type: application/json" \
+  -d '{
+    "type": "function",
+    "id": "toolu_01Vmq4gaU6tSy7ZRKVC7U2fg",
+    "function": {
+      "name": "google_search",
+      "arguments": "{\"q\":\"San Francisco news yesterday\"}"
+    }
+  }'
+
+# Step 3: Continue conversation with tool results
+curl -X POST http://localhost:8080/v1/chat/completions \
+  -H "Content-Type: application/json" \
+  -d '{
+    "provider": "openai",
+    "model": "gpt-4o-mini",
+    "messages": [
+      {"role": "user", "content": "Search for San Francisco news from yesterday"},
+      {"role": "assistant", "tool_calls": [...]},
+      {"role": "tool", "content": "...", "tool_call_id": "toolu_01Vmq4gaU6tSy7ZRKVC7U2fg"}
+    ]
+  }'
+```
+
+For detailed MCP setup and multi-turn conversation examples, see [Multi-Turn Conversations with MCP Tools](../transports/README.md#multi-turn-conversations-with-mcp-tools).
+
+### 4. Metrics
 
 **GET** `/metrics`
 

docs/logger.md

@@ -37,16 +37,16 @@ client, err := bifrost.Init(schemas.BifrostConfig{
 
 The default logger formats messages as:
 
-```
-[BIFROST-TIMESTAMP] LEVEL: message
-[BIFROST-TIMESTAMP] ERROR: (error: error_message)
+```text
+    [BIFROST-TIMESTAMP] LEVEL: message
+    [BIFROST-TIMESTAMP] ERROR: (error: error_message)
 ```
 
 Example outputs:
 
-```
-[BIFROST-2024-03-20T10:15:30Z] INFO: Initializing provider OpenAI
-[BIFROST-2024-03-20T10:15:31Z] ERROR: (error: failed to connect to provider)
+```text
+    [BIFROST-2024-03-20T10:15:30Z] INFO: Initializing provider OpenAI
+    [BIFROST-2024-03-20T10:15:31Z] ERROR: (error: failed to connect to provider)
 ```
 
 ## 3. Implementing a Custom Logger