Code Mode

AI writes Python to orchestrate tools. Reduces token usage by 50%+ when using multiple MCP servers.

This feature is only available on v1.4.0-prerelease1 and above.

Overview

Code Mode is a transformative approach to using MCP that solves a critical problem at scale:

The Problem: When you connect 8-10 MCP servers (150+ tools), every single request includes all tool definitions in the context. The LLM spends most of its budget reading tool catalogs instead of doing actual work.

The Solution: Instead of exposing 150 tools directly, Code Mode exposes just four generic tools. The LLM uses those tools to write Python code (Starlark) that orchestrates everything else in a sandbox.

The Impact

Compare a workflow across 5 MCP servers with ~100 tools:

Classic MCP Flow:

• 6 LLM turns
• 100 tools in context every turn (600 tool-definition tokens)
• All intermediate results flow through the model

Code Mode Flow:

• 3-4 LLM turns
• Only 4 tools + definitions on-demand
• Intermediate results processed in sandbox

Result: ~50% cost reduction + 30-40% faster execution

Code Mode provides four meta-tools to the AI:

1. listToolFiles - Discover available MCP servers
2. readToolFile - Load Python stub signatures on-demand
3. getToolDocs - Get detailed documentation for a specific tool
4. executeToolCode - Execute Python code with full tool bindings

When to Use Code Mode

Enable Code Mode if you have:

• ✅ 3+ MCP servers connected
• ✅ Complex multi-step workflows
• ✅ Concerned about token costs or latency
• ✅ Tools that need to interact with each other

Keep Classic MCP if you have:

• ✅ Only 1-2 small MCP servers
• ✅ Simple, direct tool calls
• ✅ Very latency-sensitive use cases (though Code Mode is usually faster)

You can mix both: Enable Code Mode for “heavy” servers (web, documents, databases) and keep small utilities as direct tools.

---

How Code Mode Works

The Four Tools

Instead of seeing 150+ tool definitions, the model sees four generic tools:

The Execution Flow

Key insight: All the complex orchestration happens inside the sandbox. The LLM only receives the final, compact result—not every intermediate step.

---

Why This Matters at Scale

Classic MCP with 5 servers (100 tools):

Turn 1: Prompt + search query + [100 tool definitions]
Turn 2: Prompt + search result + [100 tool definitions]
Turn 3: Prompt + channel list + [100 tool definitions]
Turn 4: Prompt + video list + [100 tool definitions]
Turn 5: Prompt + summaries + [100 tool definitions]
Turn 6: Prompt + doc result + [100 tool definitions]

Total: 6 LLM calls, ~600+ tokens in tool definitions alone

Code Mode with same 5 servers:

Turn 1: Prompt + 4 tools (listToolFiles, readToolFile, getToolDocs, executeToolCode)
Turn 2: Prompt + server list + 4 tools
Turn 3: Prompt + selected definitions + 4 tools + [EXECUTES CODE]
        [YouTube search, channel list, videos, summaries, doc creation all happen in sandbox]
Turn 4: Prompt + final result + 4 tools

Total: 3-4 LLM calls, ~50 tokens in tool definitions
Result: 50% cost reduction, 3-4x fewer LLM round trips

---

Enabling Code Mode

Code Mode must be enabled per MCP client. Once enabled, that client’s tools are accessed through the four meta-tools rather than exposed directly.

Best practice: Enable Code Mode for 3+ servers or any “heavy” server (web search, documents, databases).

Web UI

Enable Code Mode for a Client

1. Navigate to MCP Gateway in the sidebar
2. Click on a client row to open the configuration sheet

3. In the Basic Information section, toggle Code Mode Client to enabled
4. Click Save Changes

Once enabled:

• This client’s tools are no longer in the default tool list
• They become accessible through listToolFiles() and readToolFile()
• The AI can write code using executeToolCode() to call them

API

# When adding a new client
curl -X POST http://localhost:8080/api/mcp/client \
  -H "Content-Type: application/json" \
  -d '{
    "name": "youtube",
    "connection_type": "http",
    "connection_string": "http://localhost:3001/mcp",
    "tools_to_execute": ["*"],
    "is_code_mode_client": true
  }'

# Or update an existing client
curl -X PUT http://localhost:8080/api/mcp/client/{id} \
  -H "Content-Type: application/json" \
  -d '{
    "name": "youtube",
    "connection_type": "http",
    "connection_string": "http://localhost:3001/mcp",
    "tools_to_execute": ["*"],
    "is_code_mode_client": true
  }'

config.json

{
  "mcp": {
    "client_configs": [
      {
        "name": "youtube",
        "connection_type": "http",
        "connection_string": "http://localhost:3001/mcp",
        "tools_to_execute": ["*"],
        "is_code_mode_client": true
      },
      {
        "name": "filesystem",
        "connection_type": "stdio",
        "stdio_config": {
          "command": "npx",
          "args": ["-y", "@anthropic/mcp-filesystem"]
        },
        "tools_to_execute": ["*"],
        "is_code_mode_client": true
      }
    ]
  }
}

Go SDK Setup

mcpConfig := &schemas.MCPConfig{
    ClientConfigs: []schemas.MCPClientConfig{
        {
            Name:             "youtube",
            ConnectionType:   schemas.MCPConnectionTypeHTTP,
            ConnectionString: bifrost.Ptr("http://localhost:3001/mcp"),
            ToolsToExecute:   []string{"*"},
            IsCodeModeClient: true, // Enable code mode
        },
        {
            Name:           "filesystem",
            ConnectionType: schemas.MCPConnectionTypeSTDIO,
            StdioConfig: &schemas.MCPStdioConfig{
                Command: "npx",
                Args:    []string{"-y", "@anthropic/mcp-filesystem"},
            },
            ToolsToExecute:   []string{"*"},
            IsCodeModeClient: true, // Enable code mode
        },
    },
}

---

The Four Code Mode Tools

When Code Mode clients are connected, Bifrost automatically adds four meta-tools to every request:

1. listToolFiles

Lists all available virtual .pyi stub files for connected code mode servers.

Example output (Server-level binding):

servers/
  youtube.pyi
  filesystem.pyi

Example output (Tool-level binding):

servers/
  youtube/
    search.pyi
    get_video.pyi
  filesystem/
    read_file.pyi
    write_file.pyi

2. readToolFile

Reads a virtual .pyi file to get compact Python function signatures for tools.

Parameters:

• fileName (required): Path like servers/youtube.pyi or servers/youtube/search.pyi
• startLine (optional): 1-based starting line for partial reads
• endLine (optional): 1-based ending line for partial reads

Example output:

# youtube server tools
# Usage: youtube.tool_name(param=value)
# For detailed docs: use getToolDocs(server="youtube", tool="tool_name")

def search(query: str, maxResults: int = None) -> dict:  # Search for videos
def get_video(id: str) -> dict:  # Get video details

3. getToolDocs

Get detailed documentation for a specific tool when the compact signature from readToolFile is not sufficient.

Parameters:

• server (required): The server name (e.g., "youtube")
• tool (required): The tool name (e.g., "search")

Example output:

# ============================================================================
# Documentation for youtube.search tool
# ============================================================================
#
# USAGE INSTRUCTIONS:
# Call tools using: result = youtube.tool_name(param=value)
# No async/await needed - calls are synchronous.
#
# CRITICAL - HANDLING RESPONSES:
# Tool responses are dicts. To avoid runtime errors:
# 1. Use print(result) to inspect the response structure first
# 2. Access dict values with brackets: result["key"] NOT result.key
# 3. Use .get() for safe access: result.get("key", default)
# ============================================================================

def search(query: str, maxResults: int = None) -> dict:
    """
    Search for videos on YouTube.

    Args:
        query (str): Search query (required)
        maxResults (int): Max results to return (optional)

    Returns:
        dict: Response from the tool. Structure varies by tool.
              Use print(result) to inspect the actual structure.

    Example:
        result = youtube.search(query="...")
        print(result)  # Always inspect response first!
        value = result.get("key", default)  # Safe access
    """
    ...

4. executeToolCode

Executes Python code in a sandboxed Starlark interpreter with access to all code mode server tools.

Parameters:

• code (required): Python code to execute

Execution Environment:

• Python code runs in a Starlark interpreter (Python subset)
• All code mode servers are exposed as global objects (e.g., youtube, filesystem)
• Tool calls are synchronous - no async/await needed
• Use print() for logging (output captured in logs)
• Assign to result variable to return a value
• Tool execution timeout applies (default 30s)

Syntax notes:

• Use keyword arguments: server.tool(param="value") NOT server.tool({"param": "value"})
• Access dict values with brackets: result["key"] NOT result.key
• List comprehensions work: [x for x in items if x["active"]]

Example code:

# Search YouTube and return formatted results
results = youtube.search(query="AI news", maxResults=5)
titles = [item["snippet"]["title"] for item in results["items"]]
print("Found", len(titles), "videos")
result = {"titles": titles, "count": len(titles)}

---

Binding Levels

Code Mode supports two binding levels that control how tools are organized in the virtual file system:

Server-Level Binding (Default)

All tools from a server are grouped into a single .pyi file.

servers/
  youtube.pyi        ← Contains all youtube tools
  filesystem.pyi     ← Contains all filesystem tools

Best for:

• Servers with few tools
• When you want to see all tools at once
• Simpler discovery workflow

Tool-Level Binding

Each tool gets its own .pyi file.

servers/
  youtube/
    search.pyi
    get_video.pyi
    get_channel.pyi
  filesystem/
    read_file.pyi
    write_file.pyi
    list_directory.pyi

Best for:

• Servers with many tools
• When tools have large/complex schemas
• More focused documentation per tool

Configuring Binding Level

Binding level is a global setting that controls how Code Mode’s virtual file system is organized. It affects how the AI discovers and loads tool definitions.

Web UI

Binding level can be viewed in the MCP configuration overview:

• Server-level (default): One .pyi file per MCP server

  • Use when: 5-20 tools per server, want simple discovery
  • Example: servers/youtube.pyi contains all YouTube tools

• Tool-level: One .pyi file per individual tool

  • Use when: 30+ tools per server, want minimal context bloat
  • Example: servers/youtube/search.pyi, servers/youtube/list_channels.pyi

Both modes use the same four-tool interface (listToolFiles, readToolFile, getToolDocs, executeToolCode). The choice is purely about context efficiency per read operation.

config.json

{
  "mcp": {
    "tool_manager_config": {
      "code_mode_binding_level": "server"
    }
  }
}

Options: "server" (default) or "tool"

Go SDK

mcpConfig := &schemas.MCPConfig{
    ToolManagerConfig: &schemas.MCPToolManagerConfig{
        CodeModeBindingLevel: schemas.CodeModeBindingLevelTool, // or CodeModeBindingLevelServer
    },
    ClientConfigs: []schemas.MCPClientConfig{
        // ... clients
    },
}

---

Auto-Execution with Code Mode

Code Mode tools can be auto-executed in Agent Mode, but with additional validation:

1. The listToolFiles and readToolFile tools are always auto-executable (they’re read-only)
2. The executeToolCode tool is auto-executable only if all tool calls within the code are allowed

How Validation Works

When executeToolCode is called in agent mode:

1. Bifrost parses the Python code
2. Extracts all serverName.toolName() calls
3. Checks each call against tools_to_auto_execute for that server
4. If ALL calls are allowed → auto-execute
5. If ANY call is not allowed → return to user for approval

Example:

{
  "name": "youtube",
  "tools_to_execute": ["*"],
  "tools_to_auto_execute": ["search"],
  "is_code_mode_client": true
}

# This code WILL auto-execute (only uses search)
results = youtube.search(query="AI")
result = results

# This code will NOT auto-execute (uses delete_video which is not in auto-execute list)
youtube.delete_video(id="abc123")

---

Code Execution Environment

Available APIs

Available	Not Available
Python-like syntax	import statements
Synchronous tool calls	Classes (use dicts)
print() for logging	File I/O
Dict/List operations	Network access
List comprehensions	random, time modules

Runtime Environment Details

Engine: Starlark interpreter (Python subset)

Tool Exposure: Tools from code mode clients are exposed as global objects:

# If you have a 'youtube' code mode client with a 'search' tool
results = youtube.search(query="AI news")

Code Processing:

1. Code is validated for syntax errors
2. Tool calls are extracted and validated
3. Code executes in isolated Starlark context
4. Result variable is automatically serialized to JSON

Execution Limits:

• Default timeout: 30 seconds per tool execution
• Memory isolation: Each execution gets its own context
• No access to host file system or network
• Logs captured from print() calls

Error Handling

Bifrost provides detailed error messages with hints:

# Error: youtube is not defined
# Hints:
# - Variable or identifier 'youtube' is not defined
# - Available server keys: youtubeAPI, filesystem
# - Use one of the available server keys as the object name

Timeouts

• Default: 30 seconds per tool call
• Configure via tool_execution_timeout in tool_manager_config
• Long-running operations are interrupted with timeout error

---

Real-World Impact Comparison

Scenario: E-commerce Assistant with Multiple Services

Setup:

• 10 MCP servers (product catalog, inventory, payments, shipping, chat, analytics, docs, images, calendar, notifications)
• Average 15 tools per server = 150 total tools
• Complex multi-step task: “Find matching products, check inventory, compare prices, get shipping estimate, create quote”

Classic MCP Results

Metric	Value
LLM Turns	8-10
Tokens in Tool Defs	~2,400 per turn
Avg Request Tokens	4,000-5,000
Avg Total Cost	$3.20-4.00
Latency	18-25 seconds

Problem: Most context goes to tool definitions. Model makes redundant tool calls. Every intermediate result travels back through the LLM.

Code Mode Results

Metric	Value
LLM Turns	3-4
Tokens in Tool Defs	~100-300 per turn
Avg Request Tokens	1,500-2,000
Avg Total Cost	$1.20-1.80
Latency	8-12 seconds

Benefit: Model writes one Python script. All orchestration happens in sandbox. Only compact result returned to LLM.

---

Next Steps

Agent Mode

Combine Code Mode with auto-execution

MCP Gateway URL

Expose your tools to external clients