## Summary
- Add new `/dyad:deflake-e2e-recent-prs` command that automatically
gathers flaky E2E tests from recent PR Playwright summary comments,
ranks them by frequency, and deflakes them sequentially
- Add `claude-deflake-e2e.yml` GitHub Actions workflow that runs daily
at 2 AM PST on self-hosted macOS runners (with workflow_dispatch support
for manual triggers)
- Document the new command in `.claude/README.md`

## Test plan
- Trigger the workflow manually via `gh workflow run
claude-deflake-e2e.yml` and verify it correctly scans recent PRs for
flaky tests and attempts to deflake them
- Verify the cron schedule triggers at the expected time

🤖 Generated with [Claude Code](https://claude.com/claude-code)
<!-- devin-review-badge-begin -->

---

<a href="https://app.devin.ai/review/dyad-sh/dyad/pull/2590"
target="_blank">
  <picture>
<source media="(prefers-color-scheme: dark)"
srcset="https://static.devin.ai/assets/gh-open-in-devin-review-dark.svg?v=1">
<img
src="https://static.devin.ai/assets/gh-open-in-devin-review-light.svg?v=1"
alt="Open with Devin">
  </picture>
</a>
<!-- devin-review-badge-end -->

<!-- CURSOR_SUMMARY -->
---

> [!NOTE]
> **Medium Risk**
> Adds a scheduled GitHub Action with write permissions that can open
PRs, so misconfiguration or prompt issues could create noisy/unintended
changes; however it’s limited to CI/automation and test-only guidance.
> 
> **Overview**
> Adds a new Claude slash command, `/dyad:deflake-e2e-recent-prs`, that
scans recent PR comments for Playwright “Flaky Tests”, ranks them by
frequency, and runs deflaking steps per spec (including guidance to
disable retries via `PLAYWRIGHT_RETRIES=0`) before optionally opening a
fix PR.
> 
> Introduces a scheduled/manual GitHub Actions workflow
(`claude-deflake-e2e.yml`) that runs daily on self-hosted macOS ARM64,
installs dependencies/browsers, and invokes the new command via
`anthropics/claude-code-action`. Documentation is updated to list the
new command and the existing `/dyad:deflake-e2e` instructions are
tightened to always disable Playwright retries.
> 
> <sup>Written by [Cursor
Bugbot](https://cursor.com/dashboard?tab=bugbot) for commit
10b9158363c6b9ae9a3f3ba52ad118149fb9cbd3. This will update automatically
on new commits. Configure
[here](https://cursor.com/dashboard?tab=bugbot).</sup>
<!-- /CURSOR_SUMMARY -->

<!-- This is an auto-generated description by cubic. -->
---
## Summary by cubic
Adds a new /dyad:deflake-e2e-recent-prs command that scans recent PRs
for Playwright flake reports, ranks flaky tests, and deflakes them. Adds
a daily GitHub Action that runs it at 10:00 UTC (2 AM PST / 3 AM PDT) on
self‑hosted macOS ARM64 to keep E2E tests stable.

- **New Features**
- Command scans recent PRs (default 20), parses Playwright summary
comments from github-actions[bot], ranks by frequency, and deflakes
specs sequentially; can push fixes via /dyad:pr-push.
- New claude-deflake-e2e.yml workflow supports manual dispatch with
pr_count, sets up Node/pnpm, installs Chromium, builds the fake LLM
server, and runs the command via anthropics/claude-code-action.

- **Bug Fixes**
- Fixed spec path handling (no double .spec.ts), added gh api
--paginate, switched to generic PR search, clarified {owner}/{repo} vs
<pr_number>, and noted DST in the cron comment.
- Disabled Playwright automatic retries in all deflake steps to prevent
false passes, including debug and snapshot update commands.

<sup>Written for commit 10b9158363c6b9ae9a3f3ba52ad118149fb9cbd3.
Summary will update on new commits.</sup>

<!-- End of auto-generated description by cubic. -->

---------
Co-authored-by: Claude Opus 4.5 <noreply@anthropic.com>