oh-my-codex 0.19.0 → 0.20.0

package/dist/team/__tests__/runtime.test.js

@@ -471,27 +471,27 @@ describe('runtime', () => {
     it('resolveWorkerLaunchArgsFromEnv injects the frontier default model for executor workers', () => {
         withIsolatedDefaultModelEnv(() => {
             const args = resolveWorkerLaunchArgsFromEnv({ OMX_TEAM_WORKER_LAUNCH_ARGS: '--no-alt-screen' }, 'executor');
-            assert.deepEqual(args, ['--no-alt-screen', '--model', 'gpt-5.5']);
+            assert.deepEqual(args, ['--no-alt-screen', '--model', 'gpt-5.6-sol']);
         });
     });
     it('resolveWorkerLaunchArgsFromEnv uses medium reasoning for executor launch defaults', () => {
         withIsolatedDefaultModelEnv(() => {
             const args = resolveWorkerLaunchArgsFromEnv({ OMX_TEAM_WORKER_LAUNCH_ARGS: '--no-alt-screen' }, 'executor', undefined, resolveAgentReasoningEffort('executor'), 'codex');
-            assert.deepEqual(args, ['--no-alt-screen', '-c', 'model_reasoning_effort="medium"', '--model', 'gpt-5.5']);
+            assert.deepEqual(args, ['--no-alt-screen', '-c', 'model_reasoning_effort="medium"', '--model', 'gpt-5.6-sol']);
         });
     });
-    it('resolveWorkerLaunchArgsFromEnv keeps planner on exact gpt-5.5 medium when inherited leader is mini', () => {
+    it('resolveWorkerLaunchArgsFromEnv keeps planner on exact gpt-5.6-sol medium when inherited leader is mini', () => {
         withIsolatedDefaultModelEnv(() => {
             const args = resolveWorkerLaunchArgsFromEnv({
-                OMX_TEAM_WORKER_LAUNCH_ARGS: '--dangerously-bypass-approvals-and-sandbox --model gpt-5.4-mini',
-                [TEAM_WORKER_INHERITED_MODEL_ENV]: 'gpt-5.4-mini',
+                OMX_TEAM_WORKER_LAUNCH_ARGS: '--dangerously-bypass-approvals-and-sandbox --model gpt-5.6-terra',
+                [TEAM_WORKER_INHERITED_MODEL_ENV]: 'gpt-5.6-terra',
             }, 'planner', undefined, resolveAgentReasoningEffort('planner'), 'codex');
             assert.deepEqual(args, [
                 '--dangerously-bypass-approvals-and-sandbox',
                 '-c',
                 'model_reasoning_effort="medium"',
                 '--model',
-                'gpt-5.5',
+                'gpt-5.6-sol',
             ]);
         });
     });
@@ -507,15 +507,15 @@ describe('runtime', () => {
     it('resolveWorkerLaunchArgsFromEnv honors inherited leader model from the dedicated env path', () => {
         withIsolatedDefaultModelEnv(() => {
             const args = resolveWorkerLaunchArgsFromEnv({
-                OMX_TEAM_WORKER_LAUNCH_ARGS: '--dangerously-bypass-approvals-and-sandbox --model gpt-5.4-mini',
-                [TEAM_WORKER_INHERITED_MODEL_ENV]: 'gpt-5.4-mini',
+                OMX_TEAM_WORKER_LAUNCH_ARGS: '--dangerously-bypass-approvals-and-sandbox --model gpt-5.6-terra',
+                [TEAM_WORKER_INHERITED_MODEL_ENV]: 'gpt-5.6-terra',
             }, 'planner', undefined, resolveAgentReasoningEffort('planner'), 'codex');
             assert.deepEqual(args, [
                 '--dangerously-bypass-approvals-and-sandbox',
                 '-c',
                 'model_reasoning_effort="medium"',
                 '--model',
-                'gpt-5.5',
+                'gpt-5.6-sol',
             ]);
         });
     });
@@ -525,10 +525,10 @@ describe('runtime', () => {
     });
     it('resolveWorkerLaunchArgsFromEnv preserves explicit model in either syntax', () => {
         assert.deepEqual(resolveWorkerLaunchArgsFromEnv({ OMX_TEAM_WORKER_LAUNCH_ARGS: '--model gpt-5' }, 'explore'), ['--model', 'gpt-5']);
-        assert.deepEqual(resolveWorkerLaunchArgsFromEnv({ OMX_TEAM_WORKER_LAUNCH_ARGS: '--model=gpt-5.3' }, 'explore'), ['--model', 'gpt-5.3']);
+        assert.deepEqual(resolveWorkerLaunchArgsFromEnv({ OMX_TEAM_WORKER_LAUNCH_ARGS: '--model=gpt-5.5' }, 'explore'), ['--model', 'gpt-5.5']);
     });
     it('resolveWorkerLaunchArgsFromEnv preserves explicit env model before planner exact model', () => {
-        assert.deepEqual(resolveWorkerLaunchArgsFromEnv({ OMX_TEAM_WORKER_LAUNCH_ARGS: '--model explicit-worker-model' }, 'planner', 'gpt-5.4-mini', 'high', 'codex'), ['-c', 'model_reasoning_effort="high"', '--model', 'explicit-worker-model']);
+        assert.deepEqual(resolveWorkerLaunchArgsFromEnv({ OMX_TEAM_WORKER_LAUNCH_ARGS: '--model explicit-worker-model' }, 'planner', 'gpt-5.6-terra', 'high', 'codex'), ['-c', 'model_reasoning_effort="high"', '--model', 'explicit-worker-model']);
     });
     it('resolveWorkerLaunchArgsFromEnv uses inherited leader model for all agent types', () => {
         const args = resolveWorkerLaunchArgsFromEnv({ OMX_TEAM_WORKER_LAUNCH_ARGS: '--no-alt-screen' }, 'executor', 'gpt-4.1');
@@ -549,8 +549,8 @@ describe('runtime', () => {
             withIsolatedDefaultModelEnv(() => {
                 const lowArgs = resolveWorkerLaunchArgsFromEnv({ OMX_TEAM_WORKER_LAUNCH_ARGS: '--no-alt-screen' }, 'executor', undefined, 'low', 'codex');
                 const highArgs = resolveWorkerLaunchArgsFromEnv({ OMX_TEAM_WORKER_LAUNCH_ARGS: '--no-alt-screen' }, 'executor', undefined, 'high', 'codex');
-                assert.deepEqual(lowArgs, ['--no-alt-screen', '-c', 'model_reasoning_effort="low"', '--model', 'gpt-5.5']);
-                assert.deepEqual(highArgs, ['--no-alt-screen', '-c', 'model_reasoning_effort="high"', '--model', 'gpt-5.5']);
+                assert.deepEqual(lowArgs, ['--no-alt-screen', '-c', 'model_reasoning_effort="low"', '--model', 'gpt-5.6-sol']);
+                assert.deepEqual(highArgs, ['--no-alt-screen', '-c', 'model_reasoning_effort="high"', '--model', 'gpt-5.6-sol']);
             });
         }
         finally {
@@ -639,7 +639,7 @@ describe('runtime', () => {
             });
             const claudeArgs = resolveWorkerLaunchArgsFromEnv({ OMX_TEAM_WORKER_LAUNCH_ARGS: '--no-alt-screen --model claude-3-7-sonnet' }, 'executor', undefined, 'low', 'claude');
             const geminiArgs = resolveWorkerLaunchArgsFromEnv({ OMX_TEAM_WORKER_LAUNCH_ARGS: '--model gemini-2.0-pro' }, 'executor', undefined, 'low', 'gemini');
-            assert.deepEqual(codexArgs, ['--no-alt-screen', '-c', 'model_reasoning_effort="high"', '--model', 'gpt-5.5']);
+            assert.deepEqual(codexArgs, ['--no-alt-screen', '-c', 'model_reasoning_effort="high"', '--model', 'gpt-5.6-sol']);
             assert.deepEqual(claudeArgs, ['--no-alt-screen', '-c', 'model_reasoning_effort="low"', '--model', 'claude-3-7-sonnet']);
             assert.deepEqual(geminiArgs, ['-c', 'model_reasoning_effort="low"', '--model', 'gemini-2.0-pro']);
         }
@@ -3235,7 +3235,7 @@ sleep 5
         delete process.env.TMUX;
         process.env.OMX_TEAM_WORKER_LAUNCH_MODE = 'prompt';
         process.env.OMX_TEAM_WORKER_CLI = 'gemini';
-        process.env.OMX_TEAM_WORKER_LAUNCH_ARGS = '--model gpt-5.3-codex-spark';
+        process.env.OMX_TEAM_WORKER_LAUNCH_ARGS = '--model gpt-5.6-luna';
         process.env.OMX_GEMINI_ARGV_CAPTURE_PATH = capturePath;
         let runtime = null;
         try {
@@ -3314,7 +3314,7 @@ process.exit(0);
         delete process.env.TMUX;
         process.env.OMX_TEAM_WORKER_LAUNCH_MODE = 'prompt';
         process.env.OMX_TEAM_WORKER_CLI = 'codex';
-        process.env.OMX_TEAM_WORKER_LAUNCH_ARGS = '--model gpt-5.3-codex-spark -c model_reasoning_effort="low"';
+        process.env.OMX_TEAM_WORKER_LAUNCH_ARGS = '--model gpt-5.6-luna -c model_reasoning_effort="low"';
         try {
             await assert.rejects(() => withoutTeamWorkerEnv(() => startTeam('team-codex-explicit-launch', 'codex prompt-mode team bootstrap', 'explore', 1, [{ subject: 's', description: 'd', owner: 'worker-1' }], cwd)), /prompt_mode_codex_requires_tty/);
             assert.equal(existsSync(join(cwd, '.omx', 'state', 'team', 'team-codex-explicit-launch')), false);
@@ -3409,11 +3409,11 @@ process.on('SIGTERM', () => process.exit(0));
             const worker2Instructions = await readFile(join(cwd, '.omx', 'state', 'team', runtime.teamName, 'workers', 'worker-2', 'AGENTS.md'), 'utf-8');
             assert.match(worker1Instructions, /You are operating as the \*\*test-engineer\*\* role/);
             assert.match(worker1Instructions, /Test Engineer/);
-            assert.doesNotMatch(worker1Instructions, /exact gpt-5\.4-mini model/);
+            assert.doesNotMatch(worker1Instructions, /exact gpt-5\.6-terra model/);
             assert.match(worker2Instructions, /You are operating as the \*\*writer\*\* role/);
             assert.match(worker2Instructions, /You are Writer\./);
-            assert.doesNotMatch(worker2Instructions, /exact gpt-5\.4-mini model/);
-            assert.match(worker2Instructions, /resolved_model: gpt-5\.5/);
+            assert.doesNotMatch(worker2Instructions, /exact gpt-5\.6-terra model/);
+            assert.match(worker2Instructions, /resolved_model: gpt-5\.6-sol/);
             let worker1Args = null;
             let worker2Args = null;
             for (let attempt = 0; attempt < 50; attempt += 1) {
@@ -3432,10 +3432,10 @@ process.on('SIGTERM', () => process.exit(0));
             const worker2Joined = worker2Args.join(' ');
             assert.match(worker1Joined, /model_reasoning_effort="medium"/);
             assert.match(worker1Joined, /model_instructions_file=.*worker-1\/AGENTS\.md/);
-            assert.match(worker1Joined, /--model gpt-5\.5/);
+            assert.match(worker1Joined, /--model gpt-5\.6-sol/);
             assert.match(worker2Joined, /model_reasoning_effort="xhigh"/);
             assert.match(worker2Joined, /model_instructions_file=.*worker-2\/AGENTS\.md/);
-            assert.match(worker2Joined, /--model gpt-5\.5/);
+            assert.match(worker2Joined, /--model gpt-5\.6-sol/);
             await shutdownTeam(runtime.teamName, cwd, { force: true });
             runtime = null;
         }
@@ -3470,7 +3470,7 @@ process.on('SIGTERM', () => process.exit(0));
             await rm(cwd, { recursive: true, force: true });
         }
     });
-    it('startTeam does not apply mini guidance for exact-match negatives like gpt-5.4-mini-tuned', async () => {
+    it('startTeam does not apply mini guidance for exact-match negatives like gpt-5.6-terra-tuned', async () => {
         const cwd = await mkdtemp(join(tmpdir(), 'omx-runtime-mini-tuned-'));
         const binDir = join(cwd, 'bin');
         const fakeCodexPath = join(binDir, 'codex');
@@ -3507,7 +3507,7 @@ process.on('SIGTERM', () => process.exit(0));
         process.env.OMX_TEAM_WORKER_CLI = 'codex';
         process.env.OMX_ARGV_CAPTURE_DIR = captureDir;
         delete process.env.OMX_DEFAULT_STANDARD_MODEL;
-        process.env.OMX_TEAM_WORKER_LAUNCH_ARGS = '--model gpt-5.4-mini-tuned';
+        process.env.OMX_TEAM_WORKER_LAUNCH_ARGS = '--model gpt-5.6-terra-tuned';
         let runtime = null;
         try {
             runtime = await withMockPromptModeCodexAllowed(() => withoutTeamWorkerEnv(() => startTeam('team-mini-tuned-routing', 'mini tuned routing handoff', 'executor', 1, [
@@ -3516,9 +3516,9 @@ process.on('SIGTERM', () => process.exit(0));
             const workerInstructions = await readFile(join(cwd, '.omx', 'state', 'team', runtime.teamName, 'workers', 'worker-1', 'AGENTS.md'), 'utf-8');
             assert.match(workerInstructions, /You are operating as the \*\*writer\*\* role/);
             assert.match(workerInstructions, /You are Writer\./);
-            assert.doesNotMatch(workerInstructions, /exact gpt-5\.4-mini model/);
+            assert.doesNotMatch(workerInstructions, /exact gpt-5\.6-terra model/);
             assert.doesNotMatch(workerInstructions, /strict execution order: inspect -> plan -> act -> verify/);
-            assert.match(workerInstructions, /resolved_model: gpt-5\.4-mini-tuned/);
+            assert.match(workerInstructions, /resolved_model: gpt-5\.6-terra-tuned/);
             let workerArgs = null;
             for (let attempt = 0; attempt < 50; attempt += 1) {
                 const workerPath = join(captureDir, 'team-mini-tuned-routing__worker-1.json');
@@ -3530,7 +3530,7 @@ process.on('SIGTERM', () => process.exit(0));
             }
             assert.ok(workerArgs, 'worker argv capture file should be written');
             const workerJoined = workerArgs.join(' ');
-            assert.match(workerJoined, /--model gpt-5\.4-mini-tuned/);
+            assert.match(workerJoined, /--model gpt-5\.6-terra-tuned/);
             await shutdownTeam(runtime.teamName, cwd, { force: true });
             runtime = null;
         }
@@ -6971,7 +6971,7 @@ esac
             runtime = await withPromptModeCodexEnv(binDir, {}, () => withoutTeamWorkerEnv(() => startTeam('team-delegation-persist', 'delegation persistence test', 'executor', 1, [{ subject: 'Investigate runtime assignment', description: 'Search runtime, debug assignTask behavior, and coordinate shared handoff boundaries' }], cwd)));
             const task = await readTask(runtime.teamName, '1', cwd);
             assert.equal(task?.delegation?.mode, 'auto');
-            assert.equal(task?.delegation?.child_model, 'gpt-5.4-mini');
+            assert.equal(task?.delegation?.child_model, 'gpt-5.6-terra');
             assert.equal(task?.delegation?.required_parallel_probe, true);
             assert.equal(task?.coordination?.mode, 'coordinated');
             assert.ok(task?.coordination?.activation_reasons.includes('cross_boundary_or_handoff_language'));
@@ -7372,7 +7372,7 @@ esac
             await assert.rejects(() => assignTask('team-assign-delegation', 'worker-1', task.id, cwd), /worker_notify_failed/);
             const reread = await readTask('team-assign-delegation', task.id, cwd);
             assert.equal(reread?.delegation?.mode, 'auto');
-            assert.equal(reread?.delegation?.child_model, 'gpt-5.4-mini');
+            assert.equal(reread?.delegation?.child_model, 'gpt-5.6-terra');
             assert.equal(reread?.coordination?.mode, 'coordinated');
             assert.ok(reread?.coordination?.activation_reasons.includes('shared_file_scope'));
             assert.equal(reread?.coordination?.activation_reasons.includes('stale_snapshot_before_assignment'), false);