Agent Upstart

Cool things people are building with agents — and practical help getting the most out of them.

Playbooks · Teams tasks

Copilot code review now has an API

Written November 8, 2026

GitHub's Nov 6 changelog lets orgs manage Copilot code review by API — but GitHub Apps and fine-grained tokens hit a wall at enterprise level.

What shipped

You can now manage GitHub Copilot code review settings through the REST and GraphQL APIs, from enterprise policy down to individual repositories. The changelog is dated November 6, 2026, and the updates are available to Copilot Pro, Pro+, Max, Business, and Enterprise plans.

You could already request reviews, choose their effort level, and get information back through the REST API. Now you can also identify Copilot's reviews and comments through explicit attribution fields, see the review effort a review used when that metadata is available, and see which skills, MCP servers, and custom instructions Copilot reported as sources for a comment. That last one is the most useful of the three: when Copilot leaves a comment, you can now see what it based the comment on, instead of guessing.

The settings themselves follow your role. Enterprise administrators can enable or disable code review for all organizations, or leave the choice to organization administrators. Organization administrators can manage code review enablement, the default review effort, and the runner configuration, and settings your enterprise manages cannot be overridden. Repository administrators can manage review effort and whether reviews use custom instructions and MCP tools. The REST API accepts the same review effort levels as the web interface, and organization and repository settings responses list the levels available to you.

The auth trap

The enterprise and organization settings endpoints, including the organization runner endpoint, do not support GitHub Apps or fine-grained personal access tokens. The repository settings endpoints support both, with Administration: read permission to read settings and Administration: write permission to update them. Classic personal access tokens need repository administrator access.

That split decides how you automate this. A script that manages review settings across every repo in an organization cannot use a GitHub App at the enterprise or organization level. It needs a different credential for those endpoints, or it works repo by repo, where GitHub Apps and fine-grained tokens are accepted. Plan the credential before you write the script, not after it fails on the first call. The changelog states the restriction plainly, which is better than finding it in an error message, but it is still a wall in the middle of an API that promises org-wide management.

The review overview changes

The review overview comment is now more compact. The review effort used and any surveys move into a footer. Skipped and content-excluded files share one "Files not reviewed" section, with filenames and linked exclusion reasons. Empty sections do not appear, and long lists show an overflow count. You can also tell GitHub what is wrong with an overview: after a thumbs-down, choose Add a reason and pick the main problem, such as an approval assessment you disagree with, a comment that is too verbose, or findings that are incorrect or nitpicky.

The same changelog carries a set of smaller fixes worth knowing about. Copilot code review agent sessions now show the AI credits consumed in a review. The automatically-request-review repository rule now supports a bypass list, so exempt actors in the list will not receive an automatic review. On later reviews of the same pull request, findings unrelated to your recent changes are collapsed and then hidden. Balanced effort reviews now use three reviewer models, with another model combining their findings. Billing and preview settings are now preserved when you turn code review off and on again. Failed or timed-out reviews are retried once automatically, and Copilot can switch to another eligible model when the preferred one is unavailable. GitHub Actions, Dependabot, and apps your organization or enterprise owns can request reviews on bot-authored pull requests again when Block reviews requested with an external Copilot license is on, as long as your organization pays for the review.

What to script first

Write one script that reads the review settings for every repository in one organization and reports what sticks per repo: whether review is on, the default effort, and whether custom instructions and MCP tools are enabled. Run it read-only first. The read alone tells you how consistent your repos are, and it is the safe way to learn which credential each endpoint accepts. Only then add writes, and write the same settings to every repo in one pass, with the script reporting which repos accepted the change and which refused. Keep the enterprise-level settings out of that script until you have a credential that the enterprise endpoints accept.

Limits

Related playbooks

Sources

Back to: Small-team playbooks hub · Also: Copilot reviews: how many get addressed · Copilot Automations learn from real outcomes · Lock down Claude: IP allowlist and sign-out