TeamCity 2026.2: Pipelines and AI Features
A new era in CI/CD workflows begins.
1Pipelines Now Generally Available
With the TeamCity 2026.2 release, pipelines have moved out of the EAP phase and are now generally available. This feature is accessible for all types of projects in both TeamCity Cloud and On-Premises environments. Pipelines have narrowed the gap with classic build configurations, gaining new capabilities such as build feature support, .NET steps, and the ability to link into build chains.
2Enhanced Branch Handling
Pipelines offer significant improvements in branch management when working with YAML files stored in Git repositories. You can now select repository branches while in Edit mode and design unique workflows for each branch. Additionally, clearer warnings are provided when working with protected branches, ensuring changes made in the UI cannot be directly pushed to them.
3Run Through Failures
When setting up job dependencies, you can now enable the 'Run job even if upstream fails' option to ensure the pipeline runs to completion even in the event of errors. This feature guarantees that specific jobs, such as cleanup or notification steps, always execute, while the pipeline is still marked as failed overall.
4Promote Pipeline Runs
The Promote feature is now available not only for build configurations but also for pipelines. For example, you can promote a successful Build Docker image run to the Upload to DockerHub configuration or pipeline, allowing redeployment without rebuilding.
5Debug Individual Jobs
You can now debug individual jobs without leaving Edit mode or triggering the entire pipeline. This small but impactful addition allows for faster design of complex workflows by enabling you to fix issues as they arise rather than navigating through failure logs at the end.
6Create Pipelines Without VCS Roots
You can now create pipelines without attaching VCS roots, a feature previously available only for build configurations. By selecting the 'Without repository' option when creating a pipeline, you can design custom workflows that do not check out any remote sources.
7New MCP Tools
Your AI Agents can now use three new tools designed to work with TeamCity pipelines: teamcity_pipeline_get, teamcity_pipeline_post, and teamcity_pipeline_delete. To prevent accidents, the post and delete tools are only available in Brave mode, which must be manually enabled.
8AI Assistant Supports BYOK
AI Assistant now supports Bring Your Own Key (BYOK), allowing you to integrate any supported LLM into TeamCity. By navigating to Admin | AI Assistant and entering your developer AI API key, you can use models like Anthropic, OpenAI, or Google Gemini to inspect builds and debug your CI workflows.
Original article: The JetBrains Blog — TeamCity 2026.2: Pipelines ve AI Özellikleri