Batch Changes is better than ever

In software development, the inner loop - where developers search, read, write, and run code - has seen massive improvements thanks to AI coding assistants like Cody. But it’s the outer loop that still holds teams and larger organizations back: the manual, often bureaucratic work of reviewing, securing, and rolling out changes across thousands of repositories.

Batch Changes has long been one of Sourcegraph’s most powerful tools for scaling change across the outer loop. It helps engineering teams automate code migrations, refactors, and security updates across complex, multi-repo environments - safely and consistently.

We’re building on that foundation with a renewed focus for Batch Changes across three critical areas: security, lovability, and usability. From fine-grained access controls and signed commits, to more intuitive tooling and AI-assisted workflows, these updates and planned improvements reflect our ongoing investment in making Batch Changes even more powerful and impactful for the teams who rely on it.

Security: Confidence and control at scale

Security is at the heart of any enterprise-wide change. When rolling out updates across hundreds or thousands of repositories, teams need to ensure every change is traceable, authorized, and safe.

We’ve made meaningful investments to strengthen the security posture of Batch Changes, including:

• Fine-grained access token support: Users can now use access tokens scoped to specific repositories, a key requirement for security-conscious enterprises with strict internal policies.
• Signed commits support: Batch Changes now support both GitLab- and GitHub-style signed commits, including per-user configuration.

Lovability: Reducing friction in the outer loop

Batch Changes lives in the outer loop of development - used during moments that matter, like enforcing a new standard or shipping a security fix. While it’s not something most developers use every day, it needs to feel fast, intuitive, and reliable when they do.

This quarter, we’ve focused on making Batch Changes more approachable and less frustrating to work with:

• Search, sort, and filter improvements: You can now search for batch changes by name, and sort or filter changesets to quickly surface what’s in conflict, ready to merge, or needs attention.
• Clearer error messages: We’ve started overhauling how errors are surfaced when authoring batch specs. Previously, even minor mistakes could lead to confusing or unhelpful output and these new messages are more actionable and easier to debug.

Usability: Delivering what teams need most

As more teams rely on Batch Changes to coordinate high-stakes, large-scale code changes, the bar for usability keeps rising. Developers need tooling that not only works, but works with them - helping them move faster, avoid friction, and stay in flow.

We’ve delivered several highly requested improvements that make Batch Changes more flexible, reliable, and productive:

• Bulk changeset rebasing: Merge conflicts are a fact of life when working across many repos. Teams can now bulk rebase changesets when needed, helping keep work merge-ready in ever-changing codebases
• Batch spec templates: Save and reuse your own batch specs to streamline common workflows. This unlocks consistency across teams and reduces ramp-up time for new users.
• Flag to halt on error during preview: You can now halt execution early when errors are encountered during preview - making it easier to iterate quickly without waiting for the entire spec to run.

What’s next: Building toward an automated outer loop

Some of our most advanced customers are already experimenting with AI and agents to generate and validate large-scale code changes. We see a future where Batch Changes, Cody, and agents come together to power continuous, intelligent, large-scale code transformations. In the quarters ahead, we’re investing in:

• Smarter changeset generation
• Deeper integration with AI tools
• Better guidance for structuring large-scale changes

Batch Changes is evolving into something more than just a powerful tool - it’s becoming a strategic capability for engineering teams working in complex, messy codebases. With a strong foundation and a clear roadmap ahead, we’re committed to making large-scale code changes simpler, safer, and smarter. If you haven’t tried Batch Changes in a while, now’s the time to take another look - and if you’re already using it, we’d love your feedback as we build what’s next.

‍