docs: refine prompt-optimizer and research role prompts

docs/plans/2026-04-08-remaining-agent-prompt-refinement-report.md

@@ -0,0 +1,29 @@
+# 2026-04-08 Remaining Agent Prompt Refinement Report
+
+- Generated at: `2026-04-08T08:10:09.431008`
+- Refined agents:
+  - `prompt-optimizer`
+  - `research`
+
+## Design Choices Applied
+- autonomy mode: `high`
+- risk threshold: `medium`
+- incomplete-info policy: `assume-and-push-forward`
+
+## What Changed
+- 為 `prompt-optimizer` 補上任務澄清、假設管理、派工導向輸出格式
+- 為 `research` 補上來源可信度、事實 / 推論 / 未知分離規則
+- 為兩者補上升級線與結構化回報模板
+
+## Updated Files
+- `/home/alice/.openclaw/workspace-prompt-optimizer/AGENTS.md`
+- `/home/alice/.openclaw/workspace-prompt-optimizer/WORKFLOW.md`
+- `/home/alice/.openclaw/workspace-prompt-optimizer/REPORT_TEMPLATE.md`
+- `/home/alice/.openclaw/workspace-research/AGENTS.md`
+- `/home/alice/.openclaw/workspace-research/WORKFLOW.md`
+- `/home/alice/.openclaw/workspace-research/REPORT_TEMPLATE.md`
+
+## Notes
+- This refinement completes the core prompt set for the 5 functional agents.
+- No gateway config was applied.
+- No service was restarted.