diff --git a/e2e/create-expert/create-expert.test.ts b/e2e/create-expert/create-expert.test.ts
index a9b9372e..40961562 100644
--- a/e2e/create-expert/create-expert.test.ts
+++ b/e2e/create-expert/create-expert.test.ts
@@ -22,9 +22,10 @@ import { type CommandResult, type RunResult, withEventParsing } from "../lib/run
 const PROJECT_ROOT = path.resolve(process.cwd())
 const CLI_PATH = path.join(PROJECT_ROOT, "apps/create-expert/dist/bin/cli.js")
 // LLM API calls require extended timeout; delegation adds extra LLM round-trips.
-// The create-expert workflow involves multiple delegation round-trips which can
-// take over 2 minutes when the LLM needs to iterate on expert definitions.
-const LLM_TIMEOUT = 300_000
+// The create-expert workflow involves multiple delegation round-trips (planner →
+// definition-writer → expert-tester, with possible retries) which can exceed
+// 5 minutes in CI environments.
+const LLM_TIMEOUT = 420_000
 
 function runCreateExpert(query: string, cwd: string, timeout = LLM_TIMEOUT): Promise<RunResult> {
   const args = injectProviderArgs(["--headless", query])