I want to raise a concern about how certain bug reports are being handled, specifically those affecting Cursor’s AWS Bedrock integration.
Multiple users have reported a bug that fundamentally breaks the Bedrock integration since late October. The issue has been documented with CloudWatch logs, reproduced independently, and posted across several threads. Staff acknowledged it as “serious” on Oct 27 and said it would be passed to the team. On Nov 4, it was deprioritized with no timeline. It’s now Dec 19, almost two months later, and the bug remains.
I understand prioritization requires tradeoffs. But this particular bug has cost implications for users. It silently modifies API call parameters in ways that affect AWS billing and quota allocation. Users relying on Cursor as an interface to their own AWS infrastructure have no visibility into these modifications.
For context: Cursor’s documentation states that all Bedrock requests are “routed through our backend for final prompt building.” What it doesn’t disclose is that this backend injects API parameters that change pricing tiers and quota allocation on users’ AWS accounts. We’ve raised this with AWS to clarify expectations around accurate service representation.