Back to news
NewsJune 5, 2026· 2 min read

GitHub Copilot Agent API now lets you automate code tasks across repos

Copilot Pro, Pro+, and Max users can now start and track agent tasks via REST API in public preview. Automate refactors, migrations, and releases without manual intervention.

Our Take

This is a capability unlock for teams with sprawling codebases, but the public preview label means production readiness is still months away.

Why it matters

Practitioners managing multi-repo refactors or release cycles now have a programmatic path to offload repetitive code work to Copilot agents. The move from manual UI to API-driven automation is table-stakes for enterprise adoption.

Do this week

DevOps lead: Test the Agent tasks API against one internal migration script before the preview ends so you can benchmark time savings against your current toolchain.

Copilot Agent gets a REST API

GitHub has released the Agent tasks REST API in public preview, allowing Copilot Pro, Pro+, and Max subscribers to programmatically start and track agent tasks. The API enables users to weave Copilot cloud agent (which runs in its own development environment, makes code changes, and opens pull requests) into custom automation workflows.

The API supports three concrete use cases: fanning out refactors or migrations across multiple repositories from a single script, setting up new repositories in one click from an internal developer portal, and automatically preparing weekly releases with auto-generated release notes. Authentication is available via personal access tokens (classic and fine-grained) and OAuth tokens.

API access removes the UI bottleneck

Until now, Copilot agent was a UI-only tool. Teams managing ten or a hundred repositories faced a scaling problem: they could trigger agent tasks manually, one at a time, but could not batch them or integrate them into CI/CD pipelines. The REST API dissolves that constraint.

For large engineering teams, this shifts the labor economics. A DevOps engineer can write a script once and apply it across dozens of repos without operator intervention. That compounds for recurring tasks like dependency upgrades, style migrations, or release preparation. The agent runs in the background, validates its own changes, and opens pull requests for review.

The public preview tag matters. GitHub has not committed to SLA, performance guarantees, or API stability. Teams should expect breaking changes and potential service interruptions during the preview period.

Treat this as a pilot, not a production tool yet

If your organization manages a large monorepo or a fleet of microservices, the Agent tasks API is worth a proof-of-concept now. Pick one low-risk, repetitive task (dependency updates, linting fixes, code comment cleanup) and wire it into your internal automation layer. Measure wall-clock time and PR quality against your current workflow.

Do not migrate critical path workflows to the API until GitHub moves it out of preview and publishes stability commitments. Public preview implies backward-incompatible changes are possible. Budget for refactoring your integration if the API contract shifts.

Subscription tier constraints also matter. The API is available only to Pro, Pro+, and Max tier subscribers, not Free tier users. For teams using shared GitHub accounts or internal tooling, this adds per-seat cost if you need bot-like access to the API.

#Developer Tools#Agents#Enterprise AI
Share:
Keep reading

Related stories