opencode: subagents (agent.general) on deepseek-v4-pro via ZEN

Operator decision: main driving sessions stay glm-5.2; every task-tool subagent
runs opencode/deepseek-v4-pro (~3-5x cheaper, near-free cache hits) — the weekly
upgrade run burns most of its budget in subagents. Report generator stays glm-5.2
(launch-report.py default). Project-scoped config.

opencode.jsonc

@@ -0,0 +1,15 @@
+{
+  "$schema": "https://opencode.ai/config.json",
+  // Model split for cc-ci agent runs (operator decision 2026-08-04):
+  // - The MAIN driving session (upgrader/orchestrator/report) stays on glm-5.2 —
+  //   set via --model at launch (launch-upgrader.py / launch-report.py tier config).
+  // - Every SUBAGENT (the built-in `general` agent used by the task tool) runs on
+  //   DeepSeek V4 Pro via OpenCode ZEN: ~3-5x cheaper per token than glm-5.2 with
+  //   near-free cache hits, which is where the weekly run burns most of its budget.
+  // Project-scoped: applies to opencode sessions started in this directory tree.
+  "agent": {
+    "general": {
+      "model": "opencode/deepseek-v4-pro"
+    }
+  }
+}