fix(scripts): set the tsgo watchdog default from measured lane duration

ClawSweeper on c7a699ee82 reversed its earlier guidance: the opt-in default
adopted last iteration "deliberately preserves the indefinite tsgo hang that
this PR is meant to fix". Its objection was never that a default existed, only
that 45 minutes was unmeasured.

Measured instead of guessed: hosted tsgo lanes (check-test-types, and its core
stripes) complete in 1-2 minutes across recent successful main runs, against CI
job caps of 15-20 minutes. 30 minutes is 15-30x the observed duration, leaves
room for a far slower local host, and still bounds the 90-minute and multi-hour
wedges that motivated this PR. OPENCLAW_TSGO_TIMEOUT_MS remains the documented
override for hosts that need longer.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

docs/help/testing.md

@@ -694,11 +694,12 @@ Native dependency policy:
       after 5 minutes with no stdout or stderr output. Set
       `OPENCLAW_VITEST_NO_OUTPUT_TIMEOUT_MS=0` to disable the watchdog for
       an intentionally silent investigation.
-    - `scripts/run-tsgo.mjs` waits indefinitely by default. Set
-      `OPENCLAW_TSGO_TIMEOUT_MS` to bound a run on hosts where a wedged
-      compiler would otherwise block its caller forever; on expiry the whole
-      tsgo process tree is killed and the run fails. Values above Node's
-      timer ceiling saturate instead of collapsing to a 1ms deadline.
+    - `scripts/run-tsgo.mjs` terminates a tsgo run after 30 minutes, so a
+      wedged compiler fails loudly instead of blocking its caller forever.
+      Hosted tsgo lanes finish in 1-2 minutes; raise
+      `OPENCLAW_TSGO_TIMEOUT_MS` on a slower host. On expiry the whole tsgo
+      process tree is killed and the run fails. Values above Node's timer
+      ceiling saturate instead of collapsing to a 1ms deadline.
 
   </Accordion>
 

scripts/run-tsgo.mts

@@ -16,6 +16,12 @@ import {
   shouldSkipSparseTsgoGuardError,
 } from "./lib/tsgo-sparse-guard.mts";
 
+/**
+ * Hosted tsgo lanes finish in 1-2 minutes and their CI jobs cap at 15-20, so 30
+ * leaves headroom for a far slower local host while still bounding a wedge that
+ * would otherwise never report. Raise OPENCLAW_TSGO_TIMEOUT_MS for slower hosts.
+ */
+const DEFAULT_TSGO_TIMEOUT_MS = 30 * 60 * 1000;
 /** Node's timer ceiling: a longer delay silently becomes 1ms, so a raised override must saturate. */
 const MAX_TSGO_TIMEOUT_MS = 2_147_483_647;
 
@@ -49,14 +55,10 @@ async function main(): Promise<void> {
   }
 
   ensureRepoToolNodeModulesLink(tsgoPath);
-  // Opt-in deadline: no supported duration contract covers every host and project,
-  // so an unset value keeps the pre-existing unbounded wait rather than guessing one.
-  const timeoutMs = env.OPENCLAW_TSGO_TIMEOUT_MS?.trim()
-    ? Math.min(
-        readPositiveEnvInt("OPENCLAW_TSGO_TIMEOUT_MS", env, MAX_TSGO_TIMEOUT_MS),
-        MAX_TSGO_TIMEOUT_MS,
-      )
-    : undefined;
+  const timeoutMs = Math.min(
+    readPositiveEnvInt("OPENCLAW_TSGO_TIMEOUT_MS", env, DEFAULT_TSGO_TIMEOUT_MS),
+    MAX_TSGO_TIMEOUT_MS,
+  );
   try {
     // Managed run owns the whole tsgo process tree: on timeout it SIGKILLs the
     // process group, because a wedged checker ignores SIGTERM and would otherwise

test/scripts/run-tsgo.test.ts

@@ -277,7 +277,7 @@ describe.skipIf(process.platform === "win32")("run-tsgo watchdog", () => {
     expect(result.stderr.trim().split("\n").at(-1)).toBe("[tsgo] FAILED (exit 1)");
   }, 30_000);
 
-  it("arms no watchdog until an operator opts in", () => {
+  it("leaves a healthy run alone under the default deadline", () => {
     const cwd = createTempDir("openclaw-run-tsgo-watchdog-");
     writeFakeTsgo(cwd, "#!/bin/sh\nsleep 2\nexit 0\n");