Secondary rate limits severely impact rapid AI-assisted coding workflows

[collinsauve]

🏷️ Discussion Type

Product Feedback

Body

The current secondary rate limits are severely restricting developer velocity, particularly when using modern AI coding assistants. Tools like GitHub Copilot, Cursor, and automated workspace agents rapidly generate multiple files, trigger frequent commits, and open consecutive pull requests and pull request comments.

This is particularly noticeable in the latter case - PR comments. A very common AI skill is a the /pr-review skill that can post dozens of comments to a pull request. Run a couple of these in parallel and you will very quickly trip Github limits.

Impact

• Interrupted Workflows: Developers are forced to idly wait for cool-down periods, breaking their momentum.
• Reduced AI Utility: The primary benefit of AI coding - extreme speed and efficiency - is heavily penalized by the platform's API constraints.
• Off-Platform Usage: This forces us to work-around GitHub limits by moving our workflows off of GitHub. GitHub becomes less valuable and less sticky to the organization.

Suggested Solutions

1. AI/Copilot Tier Exemptions: Relax or dynamically scale secondary rate limits for users with active GitHub Copilot or Enterprise subscriptions.
2. Relax restrictions in certain contexts: While these type of spam prevention mechanisms are understandable in public repositories, they are less necessary in private or organizational repositories. Even more-so when operating on DRAFT PRs that the user themselves opened. What abuse are we really trying to prevent there?
3. Burst Allowances: Implement a "burst" capacity that allows a high volume of rapid creations for a short window, rather than strict per-minute caps.
4. Clearer Header Diagnostics: Provide explicit headers indicating how close a user is to hitting a secondary rate limit or when it will expire, allowing tools to self-throttle gracefully instead of failing abruptly.

[github-actions[bot]]

💬 Your Product Feedback Has Been Submitted 🎉

Thank you for taking the time to share your insights with us! Your feedback is invaluable as we build a better GitHub experience for all our users.

Here's what you can expect moving forward ⏩

• Your input will be carefully reviewed and cataloged by members of our product teams. 
  • Due to the high volume of submissions, we may not always be able to provide individual responses.
  • Rest assured, your feedback will help chart our course for product improvements.
• Other users may engage with your post, sharing their own perspectives or experiences. 
• GitHub staff may reach out for further clarification or insight. 
  • We may 'Answer' your discussion if there is a current solution, workaround, or roadmap/changelog post related to the feedback.

Where to look to see what's shipping 👀

• Read the Changelog for real-time updates on the latest GitHub features, enhancements, and calls for feedback.
• Explore our Product Roadmap, which details upcoming major releases and initiatives.

What you can do in the meantime 💻

• Upvote and comment on other user feedback Discussions that resonate with you.
• Add more information at any point! Useful details include: use cases, relevant labels, desired outcomes, and any accompanying screenshots.

As a member of the GitHub community, your participation is essential. While we can't promise that every suggestion will be implemented, we want to emphasize that your feedback is instrumental in guiding our decisions and priorities.

Thank you once again for your contribution to making GitHub even better! We're grateful for your ongoing support and collaboration in shaping the future of our platform. ⭐

[tanvishinde017]

Thanks for raising this. I've run into similar issues with AI-assisted workflows where the primary API limits aren't the problem, but secondary rate limits become the bottleneck.

I especially like the idea of better diagnostics. Exposing secondary rate limit status and reset information through headers would allow clients and AI agents to back off gracefully instead of failing unexpectedly.

The distinction between public repositories and private or draft PR workflows also seems worth considering. Activity generated by repository collaborators in their own private environments has a very different risk profile from public-facing spam scenarios.

Even if the limits themselves remain unchanged, additional observability and clearer guidance would make it much easier for tool authors to design around them.

Thanks for documenting the impact and proposing concrete solutions.