test(e2e): mock + real-endpoint mode for anthropic-tool-results-repro

Author: edelauna

apps/vscode-e2e/AGENTS.md

@@ -273,6 +273,11 @@ TEST_FILE=<your-test>.test.js pnpm --filter @roo-code/vscode-e2e test:ci
 that is what flips the test into real-endpoint mode. `TEST_FILE` is required so unrelated suites
 don't burn through the real API.
 
+**Self-skip conditions** — the test skips (rather than fails) in two situations: (1) neither
+`ANTHROPIC_API_KEY` nor `AIMOCK_URL` is set (the test has no API to talk to); (2) real-endpoint
+mode only: the model used fewer than two parallel tool calls, which means the codepath under test
+was never exercised and a green result would be inconclusive.
+
 ## Programmatic fixtures (regex matching)
 
 For requests that can't be matched by a stable substring (e.g. "starts with `<environment_details>` but not preceded by a user message"), add a programmatic fixture in `src/runTest.ts` using `mock.addFixture()` with a `RegExp` match. These are only available in replay mode and are not recorded.

apps/vscode-e2e/src/suite/anthropic-tool-results-repro.test.ts

@@ -24,7 +24,7 @@ const ANTHROPIC_API_ORIGIN = "https://api.anthropic.com"
 // across API wording tweaks.
 const TOOL_RESULT_CONTRACT_ERROR = "tool_use ids were found without tool_result blocks"
 
-// IDs the matching fixture (apps/vscode-e2e/fixtures/anthropic-tool-results-repro.json)
+// IDs the matching fixture (apps/vscode-e2e/fixtures/claude-tool-results-repro.json)
 // emits for the four parallel read_file tool calls on turn 1. Kept in sync with that file.
 const FIXTURE_TOOL_USE_IDS = [
 	"toolu_repro_190_read_1",
@@ -358,17 +358,12 @@ suite("Anthropic tool_result repro", function () {
 				}
 
 				// Mock mode: aimock serves the fixture (4 parallel read_file toolCalls on turn 1,
-				// attempt_completion on turn 2). Wait for the second /v1/messages and assert it
-				// carries 4 matching tool_result IDs.
-				await waitFor(() => taskCompleted || requests.length >= 2, {
+				// attempt_completion on turn 2). Wait for both the second /v1/messages request
+				// and task completion before asserting.
+				await waitFor(() => taskCompleted && requests.length >= 2, {
 					timeout: 120_000,
 					interval: 250,
 				})
-				await waitFor(() => requests.length >= 2, { timeout: 30_000, interval: 250 })
-
-				if (!taskCompleted) {
-					await waitFor(() => taskCompleted, { timeout: 30_000, interval: 250 })
-				}
 
 				const secondRequest = requests[1]
 				assert.ok(secondRequest, "Expected Anthropic repro to issue a second /v1/messages request")