During GameDev Day 2026, TeamCity Solutions Engineer Artem Rokhin showcased an advanced workflow designed to bridge the gap between the local development environment and cloud-based CI/CD pipelines. This methodology focuses on empowering game developers, specifically Unity engine users, to troubleshoot build errors directly from within the Rider environment without the need to constantly switch between the browser and the code editor.
Problem Scenario: Discrepancy Between Local and Production Environments
A common issue in game projects is that a build process succeeds technically in the pipeline, but visual or functional defects appear when running on staging servers. In the demonstrated model, the game worked perfectly locally but was missing the landscape when uploaded to the testing platform, even though the TeamCity build process showed a "Green Build" status.
Solution Methodology: Integrating CLI and AI
The proposed methodology relies on two integrated paths:
- Manual Inspection via TeamCity CLI: By installing the Command Line Interface (CLI) within Rider, a developer can browse the list of recent builds, inspect build logs, and verify the parameters used. In the mentioned example, searching the logs helped discover that the "Build Flavor" was set to "diagnostic" instead of the correct "game" value, which caused the loss of visual elements.
- Intelligent Inspection via MCP Protocol: This part represents a qualitative leap in the methodology, where an AI agent (such as a Claude Agent) is connected to the Rider environment via the Model Context Protocol. This connection allows the agent to access TeamCity tools, enabling the developer to ask direct questions such as: "Inspect build log number 12345 and explain why the landscape is missing." The agent pulls the data, compares pipeline settings with local project files, and provides an accurate analysis of the root cause.
What Changes Practically for Developers?
This integration changes how DevOps errors are handled in several ways:
Reduced Context Switching: Instead of copying logs from the browser and pasting them into AI tools, the agent becomes an "observer" of the pipeline within the code editor itself. Diagnostic Speed: The ability to link parameters discovered in the logs (such as Build Flavor) directly to the source code that consumes these parameters within Rider significantly accelerates the repair process. Follow-up Automation: Intelligent agents can monitor the rebuild process after a fix and notify the developer as soon as deployment is complete, providing them with a preview link.
Recommendations for Improving Pipelines
To maximize the benefit of these tools and avoid repeating such issues, it is recommended to follow these practices:
- Logging Sensitive Parameters: Ensure that all settings that change game output (such as Build Flavors) are logged in build logs to facilitate tracking.
- Early Validation: Add validation processes within build scripts that refuse to continue if the parameters used are incompatible with the target runtime environment.
- Smoke Tests: Integrate simple automated tests after the build process to ensure that scenes and basic elements are loaded before approving the deployment process.