Incredibuild Team
reading time:
Introducing the Incredibuild MCP Server
The Rise of AI-Powered Software Development
Coding agents are no longer science fiction. Tools like Cursor, GitHub Copilot, Claude Code, and Codex have fundamentally changed how developers write software. These AI-powered assistants can understand codebases, suggest implementations, refactor complex functions, and even debug tricky issues—all in real time.
With each passing month, these agents become more capable and intelligent. They’re learning to navigate complex architectures, understand nuanced design patterns, and provide increasingly sophisticated engineering solutions.
But here’s the thing: it’s all about context.
The Context Problem
Even the most intelligent coding agent is only as good as the information it has access to. An agent analyzing your C++ codebase might see your source files, understand your header dependencies, and even grasp your build system configuration. But when you’re struggling with a cryptic compilation error or a mysteriously failing build, the agent is often left guessing.
Why? Because it can’t see what actually happened during the build.
Consider a typical scenario: You’re working on a large C++ project using Incredibuild for distributed builds. Something fails. The error message is vague. You ask your coding agent for help, but without visibility into build history, timing, warnings, and error patterns, it’s operating blind.
What if your agent could see:
- Recent build results and their status
- Error and warning counts from past builds
- Build timing and duration data
That’s exactly what the Incredibuild MCP Server provides.
Enter MCP: A New Standard for Agent Capabilities
In November 2024, Anthropic released the Model Context Protocol (MCP)—an open standard that defines how AI assistants can connect to external tools and data sources. Think of MCP as a universal adapter that lets agents access information and perform actions beyond their training data.
MCP servers expose “tools” that agents can call to gather context or take action. This standardized approach means you can extend any MCP-compatible agent with new capabilities—without custom integrations for each tool.
Announcing the Incredibuild MCP Server
Today, we’re excited to release the initial version of the Incredibuild MCP Server. This open-source tool gives your coding agents direct access to your Incredibuild build history, providing the context they need to help you troubleshoot builds more effectively.
With the Incredibuild MCP Server, your agent can:
- Query recent builds and their status
- Analyze error and warning counts from builds
- Understand build timing and duration
All of this context helps your agent provide more informed, actionable suggestions when you’re debugging build issues.
Getting Started
The fastest way to get started is with Docker. The server reads from your local Incredibuild database and communicates via the standard STDIO transport that MCP clients expect.
For Cursor and Claude Desktop
Add this to your MCP configuration file (~/.cursor/mcp.json for Cursor):
JSON
{
"mcpServers": {
"incredibuild": {
"command": "docker",
"args": [
"run", "-i", "--rm",
"-v", "/path/to/your/db/folder:/data",
"ghcr.io/incredibuild-rnd/incredibuild-mcp-server:v0.1.0"
]
}
}
}On Windows, your database is typically at C:\ProgramData\Incredibuild\Manager\:
JSON
{
“mcpServers”: {
“incredibuild”: {
“command”: “docker”,
“args”: [
“run”, “-i”, “–rm”,
“-v”, “C:/ProgramData/Incredibuild/Manager:/data”,
“ghcr.io/incredibuild-rnd/incredibuild-mcp-server:v0.1.0”
]
}
}
}
Once configured, your agent will have access to two new tools:
- read_recent_builds – Query builds relative to the current time
- read_builds_in_time_range – Query builds between specific timestamps
Example Interaction
Here’s a real conversation with an agent using the Incredibuild MCP server:
You: My cmake build for Kodi just failed with 7 errors. I’ve tried rebuilding 3 times and it keeps failing. Can you help?
Agent response (actual output, unedited):
Root cause: BuildCache corruption
Found 3 consecutive failed builds with 7 errors each. All failures are from Incredibuild BuildCache issues, not CMake compilation errors.
Error breakdown:
- Restoring Outputs failed (code 4008) — Artifact unpacking failed for UPNP source files: PLTDEVICEHOST.CPP, PLTSYNCMEDIABROWSER.CPP, NPTTIME.CPP
- Verifying Restored Outputs failed (code 4009) — Hash mismatch for LIBKODI.DIR\RELEASE\.d.json
Why rebuilds keep failing: The corrupted cache entries are reused.
Solution:
Bash
ib_console /ClearCacheWithout the MCP server, you’d be staring at “7 errors” with no context. With it, the agent identified the root cause (cache corruption, not code errors), found specific error codes and files, and gave you the exact command to fix it.
How It Works
Incredibuild already stores detailed build metadata in a local SQLite database on your machine—build times, error counts, warnings, system messages, cache reports, and more. This data has always been there; you just couldn’t easily access it.
The MCP server simply exposes this database to your coding agent in a standard way. No new infrastructure, no cloud services—just a bridge between the build data you already have and the AI assistant you’re already using.
We Want Your Feedback
This is just the beginning. We’re releasing this initial version to learn from real-world usage and understand what features would be most valuable to you.
Try it out and let us know:
- How does it help with your debugging workflow?
- What additional build data would be useful?
- What other Incredibuild capabilities should be exposed via MCP?
Reach out through GitHub Issues or your usual Incredibuild support channels. We’re eager to hear your experiences and ideas for making coding agents even more powerful with Incredibuild context.
The Incredibuild MCP Server is open source and available on GitHub. Star the repo, open issues, and contribute!
Table of Contents
Shorten your builds
Incredibuild empowers your teams to be productive and focus on innovating.
