GitHub has extended the Copilot code review service so that reviews can be triggered via the public REST and GraphQL APIs, and each request can specify an effort level. At the same time the platform has moved the default effort level to Balanced for all plans, affecting new and existing repositories unless a Lite setting was explicitly chosen.
API‑driven review requests
The service now accepts review jobs through both the REST and GraphQL endpoints. Callers can include an effort_level parameter (Lite, Balanced, or any future options) when they post a review request. This makes it possible to embed Copilot reviews in CI pipelines, automation scripts, or internal tooling without manual UI interaction.
Balanced becomes the default effort level
Following the announcement on 28 August 2026, the default setting for the effort level was changed to Balanced on 28 September 2026. The change applies to all repositories and organizations that use Copilot code review, unless a user had previously selected Lite in their personal settings; that explicit choice is retained.
Scope‑based configuration
Effort level can be overridden at four hierarchical scopes:
- Enterprise – AI controls → Agents → Copilot code review
- Organization – Copilot → Code review
- Repository – Copilot → Code review
- Personal – Profile picture → Copilot settings → Copilot → Code review
Each lower level overrides the setting defined above it, allowing teams to enforce a baseline while permitting project‑specific adjustments.
Operational and security considerations
Automating review requests via the API introduces new traffic patterns that should be monitored for rate‑limit compliance and cost impact. Because the default effort level is now Balanced, review latency and token usage may differ from prior Lite defaults, which could affect pipeline timing. Teams should audit existing automation to confirm that the assumed effort level matches the new default, or explicitly set the desired level in the API call. No new authentication mechanisms are described, but standard GitHub API authentication still applies, so credential hygiene remains essential.
Related CloudNinjas coverage: AI engineering.
What This Means For Practitioners
- Update CI/CD scripts to call the Copilot review endpoints if you need automated feedback.
- Explicitly pass
effort_levelwhen the default Balanced behavior does not match your quality or performance goals. - Review the hierarchy of effort‑level settings to ensure organization‑wide policies are enforced.
- Monitor API usage and review turnaround times after the default change to detect any unexpected impact.
