[AmethystLiang]

Summary

Adds E2E test to verify that when a user creates a new GitHub issue in Orca and starts a workspace from it, the Claude agent is launched with the issue URL prefilled via the --prefill flag.

This test validates the fix for the regression where newly created issues were not properly passing their context (URL) to Claude on start.

Testing

• Added high-quality E2E test that validates the complete workflow:
  1. Create a new GitHub issue via Orca UI
  2. Click "Start workspace from issue"
  3. Verify Claude process receives --prefill and --dangerously-skip-permissions flags with the issue URL
  4. Verify the command is properly formatted (regression check for shell quoting)

Test uses mocked gh and claude CLIs to simulate GitHub interaction and capture command arguments. Cross-platform paths handled for Windows and Unix-like systems.

Security Audit

No security concerns. Test file only; uses mocked CLIs and a disposable E2E repository. Fake CLI scripts are isolated to a temporary directory and cleaned up after the test suite.

Notes

Test covers:

• Cross-platform CLI setup (.cmd batch scripts on Windows, executable shell scripts on Unix)
• GitHub issue creation flow with mocked responses
• Terminal content verification via Playwright's poll utility with appropriate timeouts
• Proper fixture composition with custom launchEnv to inject fake CLI directory into PATH

[coderabbitai]

Caution

Review failed

The pull request is closed.

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Repository UI

Review profile: CHILL

Plan: Pro

Run ID: 5ca2c5ae-7c5f-442b-90e0-0371405e36df

📥 Commits

Reviewing files that changed from the base of the PR and between 20edaad and 847f42d.

📒 Files selected for processing (1)

• tests/e2e/github-created-issue-start-prefill.spec.ts

---

📝 Walkthrough

Walkthrough

Adds a Playwright end-to-end test with fake platform-specific gh and claude executables. The test configures a disposable repository, creates a GitHub issue through the UI, starts a workspace from it, and verifies terminal output contains Claude’s expected flags and exact issue URL without malformed quoting.

🚥 Pre-merge checks | ✅ 4 | ❌ 1

❌ Failed checks (1 warning)

Check name	Status	Explanation	Resolution
Description check	⚠️ Warning	It covers summary, testing, security, and notes, but is missing the required Screenshots and AI Review Report sections.	Add a Screenshots section with 'No visual change' if applicable, and include an AI Review Report covering risks, checks, and cross-platform validation.

✅ Passed checks (4 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title is concise and accurately describes the main regression test added in the PR.
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.

✨ Finishing Touches

📝 Generate docstrings

• Create stacked PR
• Commit on current branch

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}

[coderabbitai]

Actionable comments posted: 1

🧹 Nitpick comments (2)

tests/e2e/github-created-issue-start-prefill.spec.ts (2)

164-164: 📐 Maintainability & Code Quality | 🔵 Trivial | 💤 Low value

Unnecessary RegExp for substring matching.

Playwright's getByRole name option already does case-insensitive substring matching for plain strings by default; wrapping ISSUE_TITLE in RegExp is redundant and is what triggers the static-analysis ReDoS warning. Passing the string directly avoids the warning without changing matching behavior.

-  await expect(orcaPage.getByRole('heading', { name: new RegExp(ISSUE_TITLE) })).toBeVisible({
+  await expect(orcaPage.getByRole('heading', { name: ISSUE_TITLE })).toBeVisible({

Source: Linters/SAST tools

---

170-180: 📐 Maintainability & Code Quality | 🔵 Trivial | ⚡ Quick win

Redundant terminal read after the poll succeeds.

The polled content at line 171 is discarded, then getTerminalContent is called again at line 176 for the remaining assertions — an extra IPC round-trip and a small window where content could theoretically differ between the two reads. Capture the polled value directly instead.

♻️ Proposed fix

-  await expect
-    .poll(() => getTerminalContent(orcaPage, 12_000), {
-      timeout: 30_000,
-      message: 'Claude prefill command did not reach the active terminal buffer'
-    })
-    .toContain('--prefill')
-  const terminalText = await getTerminalContent(orcaPage, 12_000)
+  let terminalText = ''
+  await expect
+    .poll(
+      async () => {
+        terminalText = await getTerminalContent(orcaPage, 12_000)
+        return terminalText
+      },
+      {
+        timeout: 30_000,
+        message: 'Claude prefill command did not reach the active terminal buffer'
+      }
+    )
+    .toContain('--prefill')

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: Repository UI

Review profile: CHILL

Plan: Pro

Run ID: c3cb8b53-d533-41b3-a7ca-8b0cba271eb6

📥 Commits

Reviewing files that changed from the base of the PR and between 0121f57 and 20edaad.

📒 Files selected for processing (1)

• tests/e2e/github-created-issue-start-prefill.spec.ts