1.7.0 Beta 1

Dyad v1.7.0 Beta 1 improves stability and performance, especially when working on large apps, adds an experimental Testing panel for AI-generated end-to-end testing, and includes fixes for preview logging, file handling, Supabase deploys, type checking, and memory-heavy workflows.

Highlights

Improved stability and performance for large apps

Dyad is now more resilient when working with large apps, long chats, large file trees, and memory-heavy development flows.

This release adds several protections against out-of-memory crashes. TypeScript checks and code explorer work now run with stronger process isolation, so large type-checking or code-indexing tasks are less likely to take down the whole app. Dyad also avoids duplicate type-check work after local-agent turns, reducing repeated memory spikes when Agent Mode already ran checks.

Large file operations are safer too. Dyad now avoids loading unnecessary file contents when the local agent lists files, prevents very large file reads from exhausting memory, streams upgrade-backup checksums instead of loading whole database files, and keeps large internal chat payloads out of renderer IPC.

Memory usage is also bounded in more places, including chat attachments, voice-to-text recordings, preview console logs, and child-process output. These changes should make Dyad more reliable during common high-memory workflows like editing large apps, reading large files, attaching files to chat, running previews, deploying Supabase functions, and using Agent Mode on bigger projects.

Testing panel (experimental)

Dyad now includes an experimental Testing panel for AI-generated end-to-end testing.

The Testing panel gives app testing a dedicated place in Dyad. You can use it to have AI help create end-to-end tests for important user flows, making it easier to check that your app still behaves as expected after changes.

This is designed for practical validation while you build. Instead of only relying on manual preview checks, the Testing panel helps turn app behavior into repeatable tests that can catch regressions earlier.

Enhancements

• Improved type-check stability by running TypeScript checks in an isolated utility process, reducing the chance that large checks crash Dyad.
• Improved code explorer stability by consolidating code explorer sessions into a byte-budgeted utility process with better memory isolation.
• Improved Agent Mode performance by skipping redundant Problems panel type-check refreshes after local-agent turns.
• Improved local agent file listing performance by avoiding unnecessary file-content reads when only file metadata is needed.
• Added safeguards against out-of-memory crashes from very large file reads in editor and agent workflows.
• Added memory limits for chat attachments, including limits for attachment count, per-file size, and combined attachment size.
• Improved voice-to-text recording handling so recording data is bounded before transcription.
• Bounded preview console memory usage so noisy logs from an app preview are truncated instead of growing without limit.
• Bounded child-process output memory and added stronger timeout handling for long-running commands.
• Reduced memory usage during upgrade backup creation by streaming checksum calculation for large files.
• Kept large internal AI message payloads out of renderer IPC, reducing memory pressure during chat and agent updates.
• Reduced Supabase Edge Function bundle concurrency to lower peak memory usage during deploys.
• Fixed Supabase partial redeploy detection for Edge Functions that import Node.js built-ins with node: specifiers.
• Added process-level memory diagnostics to session debug exports, making memory-related bug reports easier to investigate.
• Added OpenRouter app attribution headers to OpenRouter requests from Dyad.

Feedback

If you run into issues or have suggestions, please help us improve by reporting a bug or uploading a chat session.