面向 OpenClaw / Codex Agent 的决策导向 deep research skill,强调任务路由、证据可追溯、current-state verification、反证约束与可审计交付。
evaluation ai-agents deep-research decision-memo agent-skill research-workflow openclaw source-traceability
-
Updated
Apr 16, 2026 - Python