- Stable ID:
codex-user-prompt:76a27b2c15ef5701
- Source kind:
codex-session-user
- Category:
research-workflow
- Timestamp:
2026-05-10T09:09:39.331Z
- Semantic hash:
76a27b2c15ef57015186368ba3a0fff4a8c056617453f1556c982d34484ebb9f
- Public handling: selected full-text prompt with secret filtering.
Prompt Text
只读研究任务,不要修改文件。请浏览/查阅 specification/evaluation/formal-verification 相关论文和项目:AgentEval、DeepEval、CodeSpecBench、RESTestBench、VerifyLLM,以及 LLM-as-judge/evaluator generation/requirements mutation 相关顶会或强相关工作。目标是判断 DoneBench 的 DoneSpec、near-miss、self-violation 如何形成原创诊断协议,避免拼贴。输出:1) 相关工作核心技术;2) DoneBench 可借鉴但不能照搬的点;3) 可新增的原创模块候选;4) reviewer 可能说“这像 X”的地方;5) 建议写进 M6.2/M6.3 的具体内容。请附来源链接。
Reuse Notes
- EXTRACTED: This is a selected Codex prompt or automation prompt from the local Codex corpus.
- INFERRED: Future agents can reuse its structure, constraints, and acceptance criteria when creating similar Codex workflows.