When using rwx run --wait or rwx results --wait with your coding agents, you can now add a --fail-fast flag to get actionable info in the terminal as soon as the first task fails.
Use rwx results from inside a git repository to pull the results of the most recent run on your current branch. Results will be formatted as a markdown string that's also easy to feed into agentic tooling. See the docs for more.
Concurrency pools can now be paused, helpful for temporarily pausing pipelines as needed. See the docs for more information.
You can now search test results by name and file from your task's tests tab.
LLMs benefit from fresh context and up-to-date documentation, and RWX has released an official skill to enable coding agents to find relevant documentation, lint run definition files, and validate with real RWX runs. Check out the blog post for more!
You can now dispatch runs to execute custom code in response to RWX notifications. See the docs to get started.
Multi-factor authentication is available for accounts not logging in through SSO. Configure an authenticator app in your authentication settings. Require it for everybody in your organization in your organization authentication settings.
Starting in RWX CLI v3.1.0, when downloading logs and artifacts from the RWX CLI, files are now saved in a git-ignored .rwx/downloads folder in your repo, rather than in your default downloads folder. This makes it easier for LLM agents to access files it pulls, whether it's using rwx results, the MCP server or the "copy prompt" button in the UI.
You can now jump from any older run on a branch to the latest run by clicking the branch name dropdown — handy when you've pushed fixes to a failed run and want to check the results.
git/clone 2.0.0 has been released with built-in support for incremental fetching via tool caches. To upgrade, run rwx update --allow-major-version-change, rename github-access-token to github-token, and configure a tool cache on your run definition.