Repository navigation
Conversation
Signed-off-by: Codex <codex@openai.com> Rebuild this complete slice on its actual published predecessor, preserving the disclosed acceptance test and worker fixture repairs. Signed-off-by: song <liusongstep@gmail.com>
songoow
requested review from
huangruiteng,
loopx-agent and
maxliux5
as code owners
October 19, 2026 16:00
This was referenced Oct 19, 2026
This branch has not been deployed
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
目标与交付边界
初始候选有效,并不代表原研究目标已经完成;执行耗时、实际方法与失败结果需要进入下一次决策和真实后继工作。
这是 #6784 / #6838 拆分中的 C2 研究阶段与执行证据:原候选 Todo → 保留的研究后继 → 实际本地探针 → 普通 Explore 写回 → replan 与研究历程。它提供完整的本地证据链;native 提交、外部反馈与显式等待由 C3 接入。
Author Declaration
Implemented against
eac81014fd3a8842ddd420d7908657af107c6787;研究运行合同 的初始候选/研究后继与本地 probe 条款;Goal Vision 的 Structured stride control;研究 RFC §6.2、§6.3。research_loop.py::prepare_research_continuationtest_candidate_stage_leaves_research_work_and_execution_reaches_next_context[sqlite]本轮 1 passed(338.81 秒);File 参数保留未补跑。research_probe.py::run_probe;explore/execution_observation.pyexplore_result_writeback.ts、explore_turn_context.ts、replan_context.ts、stride_policy.ts变更与结果
research_loop=True使用既有 Todo 依赖保留研究后继;候选完成释放后继,准备阶段记录surface_only,已有研究不会被重新打开。evidence_conditioned_v1策略,安装本地loopx-research-probe。关闭处理时保留原入口行为。execution_observation,包括候选绑定、实际方法与变更理由。依赖与源码身份
8c212cfa1d38af23fa1a126712414ad211f43f66;27 个文件。验证与边界
8f852d440aae2875e916463236c79fa5ee984aff,与本片的相关 TS/Node 测试输入相同。test_candidate_stage_leaves_research_work_and_execution_reaches_next_context[sqlite],C2 commit17b7c4d6e1315482212a0c34e1c350ca3bce2be9/ 上述来源 tree:1 passed / 0 failed / 0 skipped(338.81 秒)。实际 Python imports 来自该 checkout,Python 3.12.14 / pytest 9.1.1;Node v24.19.0、SQLite 3.53.3,原 runtime qualification=true。源码、断言与期限未改。本轮 SQLite 原节点实际验证:初始候选完成仍保留独立研究工作;两步本地 probe 的同一观测进入 Explore、context 与 history;另一候选 SHA 的结果被拒绝且不追加;stride packet 传递实际测量并接受有证据的 replan,随后准入不同新 Todo 并实际执行五步候选,读回不同候选哈希和 stdout。它补齐 C2 独立阶段/stride/新 Todo 的完整路径,C3 的 native 反馈恢复路径另验。
这次通过对应明确的 C2 来源 tree,不是最终发布 head 的完整回归。发布组合还包含列明的前置修复与集成更新,其 packaged 验证按实际输入另列;File 参数本轮没有补跑。
视觉证据
冻结来源
eac81014fd3a8842ddd420d7908657af107c6787中保留的 public synthetic research-history 示例;本轮相关 UI 输入未变。截图展示计划/实际方法、变更理由、耗时/上限、未知搜索次数、停止原因与候选 SHA-256;它们不是新的拆分 head packaged 验证。执行观测不认证方法收益,不判定 Goal 完成。本次没有模型调用或正式 benchmark,不能据此声称首次质量或长期收益改善。前置 C1 本轮在其来源 tree 上原背压 File 单例精确诊断 1 passed(31.05 秒),未改源码、断言或期限;C1/C3 历史同节点失败保留,不能将该单例结果赋给完整 native 矩阵。
本轮发布链与集成验证
ce9f90ded5477c22bc8f14021a2377833e23dfaf;tree:bc90b589616adb2761ef2c0600df2183b8992c4f。d1dc8ebfe,treecb8b1777655d7f37e3571b5087d6cd4db286dbbd,组合本次观察的 main45d742088a3f7896a0a008761f68d2437398bcd1与上述 12 个实际 source heads。LoopX supplied-candidate 收据为in_sync、drift 为空;远端集成 head 已读回。chat_store.py指标基线,按自动合并文件实测的 2067 行结算;没有新增集成专用产品逻辑。