Google rolls out AI coding tool for GitHub repos

Google has introduced Gemini CLI GitHub Actions, a no-cost, autonomous agent for GitHub repos that works asynchronously in the background on routine coding tasks. The product is currently in beta.

Announced August 6, Gemini CLI GitHub Actions can be found on GitHub at google-github-actions/run-gemini-cli. The tool acts both as an autonomous agent for critical routine coding tasks and an on-demand collaborator that developers can delegate work to, Google said. Google said that while Gemini CLI was built for individual use in the user’s own terminal, Gemini CLI GitHub Actions was developed for team collaboration on GitHub. The tool is triggered by issues or pull requests and works asynchronously in the background, using the full context of a project to handle tasks, Google said. 

Three open-source workflows are featured as part of the tool’s launch:

Intelligent issue triage, in which Gemini CLI analyzes, labels, and prioritizes incoming issues.

Accelerated pull request reviews, in which Gemini CLI reviews pull requests for quality, style, and correctness.

On-demand collaboration, in which a developer mentions @gemini-cli in an issue or pull request to delegate tasks.

Google also drew attention to the security features of Gemini CLI GitHub Actions. The company noted that Vertex AI and Gemini Code Assist Standard and Enterprise users can use Google Cloud’s Workload Identity Federation for secure, credential-less authentication, while capabilities like command allowlisting allow users to explicitly approve every shell command the agent can execute. Finally, Gemini CLI GitHub Actions comes integrated with OpenTelemetry, an industry standard for telemetry, enabling users to stream logs and metrics to observability platforms such as Google Cloud Monitoring.