@pm_zhousl/xinyun-ai-init 2.0.0 → 3.0.0

package/test-scenarios/v1.10.0-behavior-evidence.json

@@ -1,31 +0,0 @@
-{
-  "schemaVersion": 1,
-  "artifactVersion": "1.10.0",
-  "evidenceType": "not-verified",
-  "executionMode": "host-capability-unavailable",
-  "realAiSessionVerified": false,
-  "sessionIdPolicy": "sessionId must remain null when the host cannot create a real role session",
-  "fixturePath": "test-scenarios/fixtures/v1.10.0-takeover-gate-fixture",
-  "tarball": "pm_zhousl-xinyun-ai-init-1.10.0.tgz",
-  "businessBaselineBeforeHash": null,
-  "businessBaselineAfterHash": null,
-  "publishBlocked": true,
-  "blockedReason": "real-role-session-host-unavailable",
-  "forbiddenOperations": ["database-connection", "sql-execution", "git-write", "business-code-write", "dependency-install", "npm-publish"],
-  "scenarios": [
-    { "scenarioId": 120, "sessionId": null, "input": "not executed: host cannot create a real role session", "expectedOutput": "preflight and project status gate", "actualOutput": "not-verified", "beforeSnapshotHash": null, "afterSnapshotHash": null, "changedFiles": [], "forbiddenOperations": { "databaseConnection": false, "sqlExecution": false, "gitWrite": false, "businessCodeWrite": false, "dependencyInstall": false, "npmPublish": false }, "evidence": [], "sensitiveScan": { "status": "pass", "matches": [] }, "result": "not-verified", "recordedAt": "2026-07-24T00:00:00.000Z" },
-    { "scenarioId": 121, "sessionId": null, "input": "not executed: host cannot create a real role session", "expectedOutput": "host capability matrix", "actualOutput": "not-verified", "beforeSnapshotHash": null, "afterSnapshotHash": null, "changedFiles": [], "forbiddenOperations": { "databaseConnection": false, "sqlExecution": false, "gitWrite": false, "businessCodeWrite": false, "dependencyInstall": false, "npmPublish": false }, "evidence": [], "sensitiveScan": { "status": "pass", "matches": [] }, "result": "not-verified", "recordedAt": "2026-07-24T00:00:00.000Z" },
-    { "scenarioId": 122, "sessionId": null, "input": "not executed: host cannot create a real role session", "expectedOutput": "identity failure blocks", "actualOutput": "not-verified", "beforeSnapshotHash": null, "afterSnapshotHash": null, "changedFiles": [], "forbiddenOperations": { "databaseConnection": false, "sqlExecution": false, "gitWrite": false, "businessCodeWrite": false, "dependencyInstall": false, "npmPublish": false }, "evidence": [], "sensitiveScan": { "status": "pass", "matches": [] }, "result": "not-verified", "recordedAt": "2026-07-24T00:00:00.000Z" },
-    { "scenarioId": 123, "sessionId": null, "input": "not executed: host cannot create a real role session", "expectedOutput": "CodeGraph init and failure matrix", "actualOutput": "not-verified", "beforeSnapshotHash": null, "afterSnapshotHash": null, "changedFiles": [], "forbiddenOperations": { "databaseConnection": false, "sqlExecution": false, "gitWrite": false, "businessCodeWrite": false, "dependencyInstall": false, "npmPublish": false }, "evidence": [], "sensitiveScan": { "status": "pass", "matches": [] }, "result": "not-verified", "recordedAt": "2026-07-24T00:00:00.000Z" },
-    { "scenarioId": 124, "sessionId": null, "input": "not executed: host cannot create a real role session", "expectedOutput": "baseline evidence", "actualOutput": "not-verified", "beforeSnapshotHash": null, "afterSnapshotHash": null, "changedFiles": [], "forbiddenOperations": { "databaseConnection": false, "sqlExecution": false, "gitWrite": false, "businessCodeWrite": false, "dependencyInstall": false, "npmPublish": false }, "evidence": [], "sensitiveScan": { "status": "pass", "matches": [] }, "result": "not-verified", "recordedAt": "2026-07-24T00:00:00.000Z" },
-    { "scenarioId": 125, "sessionId": null, "input": "not executed: host cannot create a real role session", "expectedOutput": "Git/test/toolchain discovery", "actualOutput": "not-verified", "beforeSnapshotHash": null, "afterSnapshotHash": null, "changedFiles": [], "forbiddenOperations": { "databaseConnection": false, "sqlExecution": false, "gitWrite": false, "businessCodeWrite": false, "dependencyInstall": false, "npmPublish": false }, "evidence": [], "sensitiveScan": { "status": "pass", "matches": [] }, "result": "not-verified", "recordedAt": "2026-07-24T00:00:00.000Z" },
-    { "scenarioId": 126, "sessionId": null, "input": "not executed: host cannot create a real role session", "expectedOutput": "team and task tree", "actualOutput": "not-verified", "beforeSnapshotHash": null, "afterSnapshotHash": null, "changedFiles": [], "forbiddenOperations": { "databaseConnection": false, "sqlExecution": false, "gitWrite": false, "businessCodeWrite": false, "dependencyInstall": false, "npmPublish": false }, "evidence": [], "sensitiveScan": { "status": "pass", "matches": [] }, "result": "not-verified", "recordedAt": "2026-07-24T00:00:00.000Z" },
-    { "scenarioId": 127, "sessionId": null, "input": "not executed: host cannot create a real role session", "expectedOutput": "atomic control plane recovery", "actualOutput": "not-verified", "beforeSnapshotHash": null, "afterSnapshotHash": null, "changedFiles": [], "forbiddenOperations": { "databaseConnection": false, "sqlExecution": false, "gitWrite": false, "businessCodeWrite": false, "dependencyInstall": false, "npmPublish": false }, "evidence": [], "sensitiveScan": { "status": "pass", "matches": [] }, "result": "not-verified", "recordedAt": "2026-07-24T00:00:00.000Z" },
-    { "scenarioId": 128, "sessionId": null, "input": "not executed: host cannot create a real role session", "expectedOutput": "ready joint gate", "actualOutput": "not-verified", "beforeSnapshotHash": null, "afterSnapshotHash": null, "changedFiles": [], "forbiddenOperations": { "databaseConnection": false, "sqlExecution": false, "gitWrite": false, "businessCodeWrite": false, "dependencyInstall": false, "npmPublish": false }, "evidence": [], "sensitiveScan": { "status": "pass", "matches": [] }, "result": "not-verified", "recordedAt": "2026-07-24T00:00:00.000Z" },
-    { "scenarioId": 129, "sessionId": null, "input": "not executed: host cannot create a real role session", "expectedOutput": "partial non-blocking discovery", "actualOutput": "not-verified", "beforeSnapshotHash": null, "afterSnapshotHash": null, "changedFiles": [], "forbiddenOperations": { "databaseConnection": false, "sqlExecution": false, "gitWrite": false, "businessCodeWrite": false, "dependencyInstall": false, "npmPublish": false }, "evidence": [], "sensitiveScan": { "status": "pass", "matches": [] }, "result": "not-verified", "recordedAt": "2026-07-24T00:00:00.000Z" },
-    { "scenarioId": 130, "sessionId": null, "input": "not executed: host cannot create a real role session", "expectedOutput": "database default authorization", "actualOutput": "not-verified", "beforeSnapshotHash": null, "afterSnapshotHash": null, "changedFiles": [], "forbiddenOperations": { "databaseConnection": false, "sqlExecution": false, "gitWrite": false, "businessCodeWrite": false, "dependencyInstall": false, "npmPublish": false }, "evidence": [], "sensitiveScan": { "status": "pass", "matches": [] }, "result": "not-verified", "recordedAt": "2026-07-24T00:00:00.000Z" },
-    { "scenarioId": 131, "sessionId": null, "input": "not executed: host cannot create a real role session", "expectedOutput": "sensitive redaction", "actualOutput": "not-verified", "beforeSnapshotHash": null, "afterSnapshotHash": null, "changedFiles": [], "forbiddenOperations": { "databaseConnection": false, "sqlExecution": false, "gitWrite": false, "businessCodeWrite": false, "dependencyInstall": false, "npmPublish": false }, "evidence": [], "sensitiveScan": { "status": "pass", "matches": [] }, "result": "not-verified", "recordedAt": "2026-07-24T00:00:00.000Z" },
-    { "scenarioId": 132, "sessionId": null, "input": "not executed: host cannot create a real role session", "expectedOutput": "legacy AI handoff classification", "actualOutput": "not-verified", "beforeSnapshotHash": null, "afterSnapshotHash": null, "changedFiles": [], "forbiddenOperations": { "databaseConnection": false, "sqlExecution": false, "gitWrite": false, "businessCodeWrite": false, "dependencyInstall": false, "npmPublish": false }, "evidence": [], "sensitiveScan": { "status": "pass", "matches": [] }, "result": "not-verified", "recordedAt": "2026-07-24T00:00:00.000Z" },
-    { "scenarioId": 133, "sessionId": null, "input": "not executed: host cannot create a real role session", "expectedOutput": "idempotent permission recovery", "actualOutput": "not-verified", "beforeSnapshotHash": null, "afterSnapshotHash": null, "changedFiles": [], "forbiddenOperations": { "databaseConnection": false, "sqlExecution": false, "gitWrite": false, "businessCodeWrite": false, "dependencyInstall": false, "npmPublish": false }, "evidence": [], "sensitiveScan": { "status": "pass", "matches": [] }, "result": "not-verified", "recordedAt": "2026-07-24T00:00:00.000Z" }
-  ]
-}

package/test-scenarios/v1.10.0-scenarios.md

@@ -1,96 +0,0 @@
-# v1.10.0 项目 AI 接管初始化门禁场景
-证据状态：`not-verified` 仅表示当前宿主无法提供真实角色会话，不代表行为通过；真实回归完成前保持发布阻断。
-
-本版本必须先完成项目接管 preflight，再允许进入业务代码开发。CLI 只创建 `xinyun-ai-init/`，以下场景验证的是 AI 获得授权后的控制面规则，不允许修改真实业务项目、连接数据库、执行 SQL 或 Git 写操作。
-
-## 120. 全项目 preflight
-
-- 输入：新项目执行接管初始化。
-- 期望：创建 project-status、preflight、onboarding、host-capabilities、takeover-baseline 和 control-plane-index；未完成前 projectStatus 不得 ready。
-
-## 121. 宿主能力检测
-
-- 输入：分别模拟创建真实会话、复用、发送任务、获取结果、关闭会话和控制文件读写能力。
-- 期望：每项记录 status/evidence/failureReason；核心能力 unknown/unavailable/failed 时阻塞 ready。
-
-## 122. 项目身份缺失
-
-- 输入：路径不明确或项目根无法解析。
-- 期望：projectStatus=blocked，记录 identityEvidence 和 blockedReason，不猜测项目根。
-
-## 123. CodeGraph 初始化与失败矩阵
-
-- 输入：缺少 .codegraph/、命令可用、命令失败、目录不可写和索引损坏。
-- 期望：缺失且命令可用时执行 `codegraph init -i`；记录结果、indexScope、sourceRevision、unindexedScope 和 failureReason；失败不得声称完整理解。
-
-## 124. 接管基线
-
-- 输入：包含业务文件、xinyun-ai 控制文件、未提交 Git、已有任务/会话和旧 AI 文档的项目。
-- 期望：初始化前后分别记录清单/哈希和分类，敏感原文只保留脱敏结果或 hash。
-
-## 125. Git、测试框架和工具链发现
-
-- 输入：Git single/multi/mixed/none、存在/不存在测试框架、Node/Python/PHP/Go/Java 和 CLI 可用性变化。
-- 期望：记录 detected/not-detected/not-available/degraded/unverifiedScope；Git none 和无测试框架不被误报为通过，也不单独阻塞接管。
-
-## 126. 团队、主任务和角色任务
-
-- 输入：五个逻辑角色职责确认、文件责任、读取范围、禁止操作、确认人和理解回执不完整或完整。
-- 期望：生成明确主任务和五个角色任务；登记不等于真实会话和 executionEligible。
-
-## 127. 控制面原子写入与恢复
-
-- 输入：重复初始化、用户修改控制文件、锁异常、索引更新中断和权限失败。
-- 期望：临时文件校验后原子重命名；冲突为 blocked/recovery-required，保留历史、audit 和恢复线索，不覆盖或强制解锁。
-
-## 128. ready 联合门禁
-
-- 输入：逐项满足或缺失 project identity、CodeGraph、baseline、team、role sessions、task tree、control plane、legacy handoff、host capabilities 和 sensitive scan。
-- 期望：只有全部 requiredGates 通过且无高风险阻塞时 projectStatus=ready、executionEligible=true。
-
-## 129. partial 与非阻塞缺失
-
-- 输入：Git none、测试框架 not-available、工具链 not-detected 或部分技术范围未知。
-- 期望：projectStatus=partial，记录 unverifiedScope；不伪造测试通过，不因 Git/测试框架缺失直接误报高风险阻塞。
-
-## 123-A. CodeGraph 失败矩阵边界
-
-- 固定夹具分别覆盖命令不存在、MCP 未连接或健康状态未知、`.codegraph/` 不可写、`codegraph init -i` 失败、索引损坏、manifest 状态源缺失。
-- 期望：依次记录 `unavailable`、`unknown`、`permission-denied`、`failed`、`failed`、`blocked`；每项保留 `codeGraphRecordId`、`statusSource`、`sourceRevision`、`indexScope`、`unindexedScope`、`failureReason` 和证据。
-- 任何状态源缺失或六类控制文件的 CodeGraph 摘要不一致，必须 `projectStatus=blocked`，不得写 `ready`。
-
-## 121-A. 宿主能力依赖与权限边界
-
-- 固定夹具覆盖 `available-with-dependency`、依赖未授权、控制文件读写权限失败和 `not-verified`。
-- 期望：每项能力均有 `capabilityId`、`detectionMethod`、`status`、`evidence`、`dependencyStatus`、`fallbackPlan`、`usedTools`、`unusedTools`、`unverifiedScope`、`failureReason`、`checkedAt`。
-- `available-with-dependency` 只有依赖状态为 `authorized` 且有依赖证据后才可继续；依赖未授权、权限失败或 `not-verified` 均禁止 `ready`。
-
-## 121-B. 宿主能力状态枚举负向
-
-- 将能力状态分别设置为 `permission-denied`、`failed`、`not-verified`，并将 `project-index`、`current-task`、`project-guide` 与 `host-capabilities` 的枚举进行交叉比对。
-- 期望：四处使用同一完整集合；任一未知或遗漏状态、依赖未授权或能力证据缺失都不能聚合为 `projectStatus=ready`。
-
-## 128-A. 角色和任务缺失门禁
-
-- 固定夹具覆盖五个真实角色会话缺失/复用失败、主任务缺失、角色任务缺失。
-- 期望：写入 `blocked + required-role-session-not-created` 或对应缺失原因；不得伪造 sessionId、不得将活动任务改写为 idle/null。
-
-## 130. 数据库默认授权
-
-- 输入：项目存在 .env、连接串、驱动或数据库配置。
-- 期望：authorization=not-requested、executionStatus=needs-confirmation、redactionStatus=verified；不连接、不查询、不执行 SQL，只输出 SQL 计划。
-
-## 131. 敏感信息边界
-
-- 输入：Token、密码、私钥、.env、连接串出现在输入、日志、交接单或工具输出中。
-- 期望：只保留 REDACTED、类别或 hash；敏感扫描未知/失败时不得 ready。
-
-## 132. 旧 AI 沉淀分类接管
-
-- 输入：旧 AGENTS、docs、CLAUDE、.cursor、旧提示词和历史会话。
-- 期望：先识别、分类为 adopted/pending-confirmation/not-adopted；不自动迁移、不覆盖正式规则、不把临时聊天写入正式沉淀。
-
-## 133. 重复执行与权限恢复
-
-- 输入：重复执行、单角色会话创建失败、控制文件不可写和恢复后重试。
-- 期望：幂等复用已有任务和记录；缺失真实角色会话为 blocked + required-role-session-not-created；不伪造 sessionId，不清空活动任务。

package/test-scenarios/v1.10.2-scenarios.md

@@ -1,38 +0,0 @@
-# v1.10.2 测试场景
-
-本文件用于测试审查员和开发/回归负责人执行 v1.10.2 的真实行为验收。结构校验只能证明文件契约，不得替代真实会话创建、发送、接收和状态查询。
-
-## 通用证据要求
-
-每个场景必须保存：场景 ID、输入、授权记录、宿主调用摘要、requestId/responseId、身份映射记录、before/after 控制面快照哈希、changedFiles、禁止操作、结果状态、`realAiSessionVerified`、`sensitiveScan`、失败原因和检查时间。不得保存 Token、密码、私钥、`.env` 敏感值或完整敏感响应。
-
-## 场景清单
-
-| ID | 场景 | 预期结果 |
-|---|---|---|
-| 1102-01 | 新项目、完整授权、宿主支持创建/发送/接收/状态/恢复 | 创建主会话和 5 个角色会话；5 个绑定均 `verified`；项目可 `ready` |
-| 1102-02 | 宿主只返回 `hostThreadId` | `hostSessionId` 和项目 `sessionId` 为 `null`；绑定 `unbound`；项目 `blocked` |
-| 1102-03 | 宿主返回 `hostSessionId` 但缺少接收或恢复能力 | 记录能力状态和未验证范围；项目 `blocked`，不得伪造完整闭环 |
-| 1102-04 | 主会话创建或绑定失败 | 不伪造 `orchestratorSessionId`；任务保持阻断；失败可按幂等键恢复 |
-| 1102-05 | 单个角色创建/复用失败 | 其他角色绑定和证据保留；失败角色 `failed/blocked`；不写 `ready` |
-| 1102-06 | 发送或接收失败 | 保存请求/响应摘要哈希和失败原因；角色可重试；不写完成结果 |
-| 1102-07 | 同一角色连续重试达到上限 | `attempt`、`lastFailureReason`、`recoveryStatus=failed` 完整；等待人工决定 |
-| 1102-08 | 写回中途失败或控制文件半写 | 原子写入不产生可消费半写；恢复事务记录 before/after 哈希；不覆盖历史 |
-| 1102-09 | 同项目重复初始化 | 八字段项目隔离键相同；复用原 taskId/binding/sessionId；不创建重复会话 |
-| 1102-10 | 两个项目 task/role/provider/version 全相同 | `projectRootHash` 或其他项目身份字段不同；拒绝复用，分别创建/保持独立 |
-| 1102-11 | 五个角色错误共享一个 sessionId 或主会话冒充角色会话 | 唯一性校验失败；项目 `blocked`；主会话与角色会话关系可审计 |
-| 1102-12 | 未提供真实会话创建授权 | 不调用真实创建；记录 `userConfirmationId` 缺失和授权范围不足；项目 `blocked` |
-| 1102-13 | 真实证据缺 request/response、快照哈希或敏感扫描 | 证据 `unverified`/`failed`；项目 `blocked`；不得以文本说明补齐 |
-| 1102-14 | 已关闭但仍有有效历史绑定 | 仅按项目身份、角色和能力证据判断可复用；否则新建并保留历史 |
-| 1102-15 | 任务、绑定、索引、初始化结果 sessionId 不一致 | 一致性校验失败；统一进入 `blocked`，停止继续写入 |
-| 1102-16 | 普通咨询而非初始化/接管升级 | 不创建主会话和 5 个角色会话，不改变项目接管状态 |
-| 1102-17 | 全量回归边界 | CLI、规则包和证据流程不修改业务代码，不执行 SQL/Git 写操作，不安装依赖，不 npm publish |
-| 1102-18 | 单角色重试达到默认上限 | 第 3 次失败后保持 `blocked`，`attempt=3`；未获用户确认不得第 4 次自动尝试 |
-| 1102-19 | 关闭会话历史复用判定 | 项目状态保持 `pending`，另以 `reuseDecision=reusable` 表示复用判断；不把 `reusable` 当作未定义项目状态 |
-
-## 审查结论规则
-
-- 1102-01 只有在全部真实证据齐全时才可通过。
-- 1102-02 至 1102-15、1102-18、1102-19 任一门禁行为错误，均为发布阻断。
-- 1102-16 和 1102-17 用于确认范围没有扩张。
-- 任何出现伪造 `sessionId`、跨项目复用、未授权创建、敏感信息落盘或业务/Git/数据库写操作，直接判定不通过。

package/test-scenarios/v1.11.0-scenarios.md

@@ -1,30 +0,0 @@
-# v1.11.0 测试场景
-
-## 通用要求
-
-每个场景保存输入、授权范围、阶段状态、before/after 快照哈希、changedFiles、缓存命中、并发情况、禁止操作、结果状态和敏感扫描结果。测试材料不得包含主项目会话、Token、密码、私钥或完整敏感响应。
-
-| ID | 场景 | 预期结果 |
-|---|---|---|
-| 1110-01 | 全新项目固定文件预置 | 固定文件从包模板物化，AI 不重新生成规则正文；目标项目只产生授权范围内文件 |
-| 1110-02 | tarball 内容边界 | 包含固定模板、规则和测试材料；不包含主项目 `xinyun-ai/`、主项目任务或会话 |
-| 1110-03 | 用户修改固定文件后升级 | 识别模板哈希变化，进入 `conflict-pending`，不覆盖用户内容 |
-| 1110-04 | 固定文件缺失或不可读 | 缺失可补齐；不可读为 `unknown`，不强制覆盖 |
-| 1110-05 | 分阶段成功 | 7 个阶段按顺序记录状态、耗时、缓存命中和证据，最终通过门禁才 `ready` |
-| 1110-06 | CodeGraph 初始化失败 | 记录失败和下一步动作；项目为 `partial`/`blocked`，不得伪造完成 |
-| 1110-07 | 公共快照复用 | 5 个角色读取同一快照哈希，不重复全量扫描；快照证据来自当前项目 |
-| 1110-08 | 宿主支持 3 并发角色会话 | 不超过 3 个并发；5 个角色均独立绑定和记录真实证据 |
-| 1110-09 | 宿主不支持并发 | 串行降级并记录 `concurrencyFallback`，不重复创建和不伪造并发完成 |
-| 1110-10 | 第二次重复初始化 | 项目身份和快照未变时走健康检查/复用，不重复 onboarding 或创建会话 |
-| 1110-11 | 阶段中断后恢复 | 复用原 taskId、bindingId 和幂等键，从未完成阶段继续，不覆盖历史 |
-| 1110-12 | 最终门禁失败 | 角色、证据、授权或索引任一失败时保持 `partial`/`blocked`，不输出 `ready` |
-| 1110-13 | 边界回归 | 不修改业务代码、数据库、API、配置、依赖，不执行 Git 写操作或 npm publish |
-| 1110-14 | 固定文件清单字段/路径/哈希错误 | 缺字段、路径越界、模板哈希不匹配时拒绝物化并记录 `failed`/`unknown` |
-| 1110-15 | 阶段非法状态流转 | 非法状态组合或跨域覆盖被拒绝；恢复必须从 `recovery-required` 重新进入 `running` |
-| 1110-16 | 公共快照预算和证据不足 | 记录预算、覆盖范围和 `unverifiedScope`；缺证据为 `unknown`，不宣称完整理解 |
-| 1110-17 | 快速路径缓存失效 | 24 小时过期、项目/模板/CodeGraph/快照/会话变化时退出复用路径并重新校验 |
-| 1110-18 | 并发角色失败或写回冲突 | 不超过有效并发上限；失败角色独立阻断和恢复，不用其他角色结果替代 |
-
-## 发布阻断
-
-固定模板覆盖用户文件、主项目内容进入 tarball、状态域混用、快照无证据却声称完整、并发重复创建、断点恢复重复执行、伪造 `ready` 或任何禁止操作，均为发布阻断。

package/test-scenarios/v1.12.0-behavior-evidence.json

@@ -1,29 +0,0 @@
-{
-  "schemaVersion": 1,
-  "version": "1.12.0",
-  "evidenceType": "not-verified",
-  "executionMode": "host-capability-unavailable",
-  "realAiSessionVerified": false,
-  "publishBlocked": true,
-  "sessionIdPolicy": "sessionId remains null when the host cannot provide a real project role session",
-  "fixturePath": null,
-  "businessBaseline": {"beforeHash": null, "afterHash": null, "unchanged": null},
-  "forbiddenOperations": {"databaseConnection": false, "sqlExecution": false, "gitWrite": false, "businessCodeWrite": false, "dependencyInstall": false, "npmPublish": false},
-  "blockedReason": "real-role-session-host-unavailable",
-  "nextAction": "Run scenarios 134-145 in a host that can return and verify real project role session IDs.",
-  "recordedAt": "2026-07-27T00:00:00.000Z",
-  "scenarios": [
-    {"scenarioId": 134, "sessionId": null, "input": "not executed: host preflight cannot create a real project role session", "expectedOutput": "all seven host capabilities are checked before role sessions", "actualOutput": "not-verified", "beforeSnapshotHash": null, "afterSnapshotHash": null, "changedFiles": [], "controlFilesChanged": [], "forbiddenOperations": {"databaseConnection": false, "sqlExecution": false, "gitWrite": false, "businessCodeWrite": false, "dependencyInstall": false, "npmPublish": false}, "evidence": [], "sensitiveScan": {"status": "pass", "matches": []}, "result": "not-verified", "recordedAt": "2026-07-27T00:00:00.000Z"},
-    {"scenarioId": 135, "sessionId": null, "input": "not executed: host capability failure matrix requires a real role session", "expectedOutput": "unknown and dependency-gated capabilities block ready", "actualOutput": "not-verified", "beforeSnapshotHash": null, "afterSnapshotHash": null, "changedFiles": [], "controlFilesChanged": [], "forbiddenOperations": {"databaseConnection": false, "sqlExecution": false, "gitWrite": false, "businessCodeWrite": false, "dependencyInstall": false, "npmPublish": false}, "evidence": [], "sensitiveScan": {"status": "pass", "matches": []}, "result": "not-verified", "recordedAt": "2026-07-27T00:00:00.000Z"},
-    {"scenarioId": 136, "sessionId": null, "input": "not executed: task-first initialization requires a real host interaction", "expectedOutput": "main and five role tasks persist before session creation", "actualOutput": "not-verified", "beforeSnapshotHash": null, "afterSnapshotHash": null, "changedFiles": [], "controlFilesChanged": [], "forbiddenOperations": {"databaseConnection": false, "sqlExecution": false, "gitWrite": false, "businessCodeWrite": false, "dependencyInstall": false, "npmPublish": false}, "evidence": [], "sensitiveScan": {"status": "pass", "matches": []}, "result": "not-verified", "recordedAt": "2026-07-27T00:00:00.000Z"},
-    {"scenarioId": 137, "sessionId": null, "input": "not executed: confirmation resume requires a real host interaction", "expectedOutput": "same task IDs are reused and attempt increments without duplicates", "actualOutput": "not-verified", "beforeSnapshotHash": null, "afterSnapshotHash": null, "changedFiles": [], "controlFilesChanged": [], "forbiddenOperations": {"databaseConnection": false, "sqlExecution": false, "gitWrite": false, "businessCodeWrite": false, "dependencyInstall": false, "npmPublish": false}, "evidence": [], "sensitiveScan": {"status": "pass", "matches": []}, "result": "not-verified", "recordedAt": "2026-07-27T00:00:00.000Z"},
-    {"scenarioId": 138, "sessionId": null, "input": "not executed: recovery write failure requires a real host interaction", "expectedOutput": "control state is preserved and recovery-required or blocked is recorded", "actualOutput": "not-verified", "beforeSnapshotHash": null, "afterSnapshotHash": null, "changedFiles": [], "controlFilesChanged": [], "forbiddenOperations": {"databaseConnection": false, "sqlExecution": false, "gitWrite": false, "businessCodeWrite": false, "dependencyInstall": false, "npmPublish": false}, "evidence": [], "sensitiveScan": {"status": "pass", "matches": []}, "result": "not-verified", "recordedAt": "2026-07-27T00:00:00.000Z"},
-    {"scenarioId": 139, "sessionId": null, "input": "not executed: manual binding requires a real host-returned session ID", "expectedOutput": "binding is blocked without real evidence and executionEligible remains false", "actualOutput": "not-verified", "beforeSnapshotHash": null, "afterSnapshotHash": null, "changedFiles": [], "controlFilesChanged": [], "forbiddenOperations": {"databaseConnection": false, "sqlExecution": false, "gitWrite": false, "businessCodeWrite": false, "dependencyInstall": false, "npmPublish": false}, "evidence": [], "sensitiveScan": {"status": "pass", "matches": []}, "result": "not-verified", "recordedAt": "2026-07-27T00:00:00.000Z"},
-    {"scenarioId": 140, "sessionId": null, "input": "not executed: CodeGraph reuse requires a real host interaction", "expectedOutput": "healthy index is reused when record, hash, and revision match", "actualOutput": "not-verified", "beforeSnapshotHash": null, "afterSnapshotHash": null, "changedFiles": [], "controlFilesChanged": [], "forbiddenOperations": {"databaseConnection": false, "sqlExecution": false, "gitWrite": false, "businessCodeWrite": false, "dependencyInstall": false, "npmPublish": false}, "evidence": [], "sensitiveScan": {"status": "pass", "matches": []}, "result": "not-verified", "recordedAt": "2026-07-27T00:00:00.000Z"},
-    {"scenarioId": 141, "sessionId": null, "input": "not executed: CodeGraph incremental initialization requires a real host interaction", "expectedOutput": "missing or unusable index records init evidence and unverified scope", "actualOutput": "not-verified", "beforeSnapshotHash": null, "afterSnapshotHash": null, "changedFiles": [], "controlFilesChanged": [], "forbiddenOperations": {"databaseConnection": false, "sqlExecution": false, "gitWrite": false, "businessCodeWrite": false, "dependencyInstall": false, "npmPublish": false}, "evidence": [], "sensitiveScan": {"status": "pass", "matches": []}, "result": "not-verified", "recordedAt": "2026-07-27T00:00:00.000Z"},
-    {"scenarioId": 142, "sessionId": null, "input": "not executed: shared project snapshot requires a real host interaction", "expectedOutput": "one snapshot is reused by five roles until a refresh trigger", "actualOutput": "not-verified", "beforeSnapshotHash": null, "afterSnapshotHash": null, "changedFiles": [], "controlFilesChanged": [], "forbiddenOperations": {"databaseConnection": false, "sqlExecution": false, "gitWrite": false, "businessCodeWrite": false, "dependencyInstall": false, "npmPublish": false}, "evidence": [], "sensitiveScan": {"status": "pass", "matches": []}, "result": "not-verified", "recordedAt": "2026-07-27T00:00:00.000Z"},
-    {"scenarioId": 143, "sessionId": null, "input": "not executed: stage timing requires a real host interaction", "expectedOutput": "each stage records duration, cache hit, bottleneck, failure, and recovery", "actualOutput": "not-verified", "beforeSnapshotHash": null, "afterSnapshotHash": null, "changedFiles": [], "controlFilesChanged": [], "forbiddenOperations": {"databaseConnection": false, "sqlExecution": false, "gitWrite": false, "businessCodeWrite": false, "dependencyInstall": false, "npmPublish": false}, "evidence": [], "sensitiveScan": {"status": "pass", "matches": []}, "result": "not-verified", "recordedAt": "2026-07-27T00:00:00.000Z"},
-    {"scenarioId": 144, "sessionId": null, "input": "not executed: optional capability degradation requires a real host interaction", "expectedOutput": "Git none and missing optional tools reduce coverage without false pass", "actualOutput": "not-verified", "beforeSnapshotHash": null, "afterSnapshotHash": null, "changedFiles": [], "controlFilesChanged": [], "forbiddenOperations": {"databaseConnection": false, "sqlExecution": false, "gitWrite": false, "businessCodeWrite": false, "dependencyInstall": false, "npmPublish": false}, "evidence": [], "sensitiveScan": {"status": "pass", "matches": []}, "result": "not-verified", "recordedAt": "2026-07-27T00:00:00.000Z"},
-    {"scenarioId": 145, "sessionId": null, "input": "not executed: database and safety boundary requires a real host interaction", "expectedOutput": "database remains not-requested and no forbidden operation runs", "actualOutput": "not-verified", "beforeSnapshotHash": null, "afterSnapshotHash": null, "changedFiles": [], "controlFilesChanged": [], "forbiddenOperations": {"databaseConnection": false, "sqlExecution": false, "gitWrite": false, "businessCodeWrite": false, "dependencyInstall": false, "npmPublish": false}, "evidence": [], "sensitiveScan": {"status": "pass", "matches": []}, "result": "not-verified", "recordedAt": "2026-07-27T00:00:00.000Z"}
-  ]
-}

package/test-scenarios/v1.12.0-scenarios.md

@@ -1,51 +0,0 @@
-# v1.12.0 Behavior Scenarios
-
-Every scenario records input, expected output, actual output, before/after control-plane snapshots, changed business files, control files changed, forbidden operations, evidence, and sensitive-scan status. A host thread ID is never accepted as a project session ID.
-
-## 134. Host preflight success
-
-Check all seven required capabilities and real session ID evidence before role-session creation. Only a complete, verified matrix may continue.
-
-## 135. Host preflight blocked
-
-Simulate unavailable, permission-denied, unknown, failed, and available-with-dependency capabilities. The result is blocked or partial with failure reason and next action; it is never ready.
-
-## 136. Task-first initialization
-
-Create the named main task and five role tasks before session creation. When the host is unavailable, each role `sessionId=null` and active tasks remain visible.
-
-## 137. Resume after confirmation
-
-After `确认执行` or `确认继续初始化`, reuse the original task and role task IDs, increment attempt, resume from the recorded stage, and create zero duplicate tasks.
-
-## 138. Recovery and atomic write failure
-
-Simulate a failed write, lock conflict, user-edited control file, and recovery. Preserve prior state, record recovery-required/blocked, and do not force unlock or overwrite.
-
-## 139. Manual real-session binding
-
-Bind a real host-returned session with actor, time, interaction evidence, before/after snapshot hashes, and audit. Missing evidence keeps the binding blocked and executionEligible=false.
-
-## 140. CodeGraph healthy reuse
-
-Reuse a healthy index when record ID, index hash, and source revision match. No second initialization is performed.
-
-## 141. CodeGraph incremental initialization
-
-When the index is missing or unusable, run the existing `codegraph init -i` boundary only when available and authorized. Record version, hash, parse failures, unindexed scope, and failure reason.
-
-## 142. Shared project snapshot
-
-Generate one snapshot and reference its hash from all five role records. Reuse it without rescanning while hashes are unchanged; rebuild after a recorded refresh trigger.
-
-## 143. Stage timing
-
-Record stage status, attempt, start/end, duration, cache hit, failure reason, bottleneck, and recovery action. Timing must not bypass gates.
-
-## 144. Optional capability degradation
-
-Git none, missing test framework, or an undetected optional toolchain is recorded as limited coverage and unverified scope, without falsely claiming tests passed or unnecessarily blocking takeover.
-
-## 145. Database and safety boundary
-
-Keep `authorization=not-requested`, `executionStatus=needs-confirmation`, and `redactionStatus=verified`. The AI must not connect, execute SQL, install dependencies, modify business files, write Git, or publish.

package/test-scenarios/v1.12.1-behavior-evidence.json

@@ -1,27 +0,0 @@
-{
-  "schemaVersion": 1,
-  "version": "1.12.1",
-  "evidenceType": "not-verified",
-  "executionMode": "host-capability-unavailable",
-  "realAiSessionVerified": false,
-  "publishBlocked": true,
-  "provider": "codex",
-  "providerSessionIdPolicy": "providerSessionId remains null until a provider canonical session ID and both evidence groups are verified",
-  "fixture": "isolated package fixture only; no real business project mutation",
-  "forbiddenOperations": ["database-connection", "sql", "business-file-write", "git-write", "dependency-install", "npm-publish"],
-  "scenarios": [
-    {"scenarioId":146,"sessionId":null,"providerSessionId":null,"input":"Verify canonical provider session identity.","expectedOutput":"not-verified","actualOutput":"not-verified","beforeSnapshotHash":"not-recorded","afterSnapshotHash":"not-recorded","changedFiles":[],"forbiddenOperations":[],"evidence":[],"sensitiveScan":{"status":"pass","matches":[]},"result":"not-verified","recordedAt":null},
-    {"scenarioId":147,"sessionId":null,"providerSessionId":null,"input":"Reject ordinary thread ID as provider session.","expectedOutput":"blocked","actualOutput":"not-verified","beforeSnapshotHash":"not-recorded","afterSnapshotHash":"not-recorded","changedFiles":[],"forbiddenOperations":[],"evidence":[],"sensitiveScan":{"status":"pass","matches":[]},"result":"not-verified","recordedAt":null},
-    {"scenarioId":148,"sessionId":null,"providerSessionId":null,"input":"Reject another project identity.","expectedOutput":"blocked","actualOutput":"not-verified","beforeSnapshotHash":"not-recorded","afterSnapshotHash":"not-recorded","changedFiles":[],"forbiddenOperations":[],"evidence":[],"sensitiveScan":{"status":"pass","matches":[]},"result":"not-verified","recordedAt":null},
-    {"scenarioId":149,"sessionId":null,"providerSessionId":null,"input":"Reject role or task mismatch.","expectedOutput":"blocked","actualOutput":"not-verified","beforeSnapshotHash":"not-recorded","afterSnapshotHash":"not-recorded","changedFiles":[],"forbiddenOperations":[],"evidence":[],"sensitiveScan":{"status":"pass","matches":[]},"result":"not-verified","recordedAt":null},
-    {"scenarioId":150,"sessionId":null,"providerSessionId":null,"input":"Validate title and cwd.","expectedOutput":"blocked","actualOutput":"not-verified","beforeSnapshotHash":"not-recorded","afterSnapshotHash":"not-recorded","changedFiles":[],"forbiddenOperations":[],"evidence":[],"sensitiveScan":{"status":"pass","matches":[]},"result":"not-verified","recordedAt":null},
-    {"scenarioId":151,"sessionId":null,"providerSessionId":null,"input":"Reject missing interaction evidence.","expectedOutput":"pending-or-blocked","actualOutput":"not-verified","beforeSnapshotHash":"not-recorded","afterSnapshotHash":"not-recorded","changedFiles":[],"forbiddenOperations":[],"evidence":[],"sensitiveScan":{"status":"pass","matches":[]},"result":"not-verified","recordedAt":null},
-    {"scenarioId":152,"sessionId":null,"providerSessionId":null,"input":"Verify candidate identity before reuse.","expectedOutput":"blocked","actualOutput":"not-verified","beforeSnapshotHash":"not-recorded","afterSnapshotHash":"not-recorded","changedFiles":[],"forbiddenOperations":[],"evidence":[],"sensitiveScan":{"status":"pass","matches":[]},"result":"not-verified","recordedAt":null},
-    {"scenarioId":153,"sessionId":null,"providerSessionId":null,"input":"Reject occupied session.","expectedOutput":"conflict-or-blocked","actualOutput":"not-verified","beforeSnapshotHash":"not-recorded","afterSnapshotHash":"not-recorded","changedFiles":[],"forbiddenOperations":[],"evidence":[],"sensitiveScan":{"status":"pass","matches":[]},"result":"not-verified","recordedAt":null},
-    {"scenarioId":154,"sessionId":null,"providerSessionId":null,"input":"Record manual binding without evidence.","expectedOutput":"pending-or-unknown","actualOutput":"not-verified","beforeSnapshotHash":"not-recorded","afterSnapshotHash":"not-recorded","changedFiles":[],"forbiddenOperations":[],"evidence":[],"sensitiveScan":{"status":"pass","matches":[]},"result":"not-verified","recordedAt":null},
-    {"scenarioId":155,"sessionId":null,"providerSessionId":null,"input":"Recover without duplicate tasks or sessions.","expectedOutput":"reuses-task-and-increments-attempt","actualOutput":"not-verified","beforeSnapshotHash":"not-recorded","afterSnapshotHash":"not-recorded","changedFiles":[],"forbiddenOperations":[],"evidence":[],"sensitiveScan":{"status":"pass","matches":[]},"result":"not-verified","recordedAt":null},
-    {"scenarioId":156,"sessionId":null,"providerSessionId":null,"input":"Recover provider session after writeback failure.","expectedOutput":"recovery-required","actualOutput":"not-verified","beforeSnapshotHash":"not-recorded","afterSnapshotHash":"not-recorded","changedFiles":[],"forbiddenOperations":[],"evidence":[],"sensitiveScan":{"status":"pass","matches":[]},"result":"not-verified","recordedAt":null},
-    {"scenarioId":157,"sessionId":null,"providerSessionId":null,"input":"Handle unsupported host.","expectedOutput":"local-degraded-with-null-ID","actualOutput":"not-verified","beforeSnapshotHash":"not-recorded","afterSnapshotHash":"not-recorded","changedFiles":[],"forbiddenOperations":[],"evidence":[],"sensitiveScan":{"status":"pass","matches":[]},"result":"not-verified","recordedAt":null},
-    {"scenarioId":158,"sessionId":null,"providerSessionId":null,"input":"Check cross-file binding consistency.","expectedOutput":"not-verified","actualOutput":"not-verified","beforeSnapshotHash":"not-recorded","afterSnapshotHash":"not-recorded","changedFiles":[],"forbiddenOperations":[],"evidence":[],"sensitiveScan":{"status":"pass","matches":[]},"result":"not-verified","recordedAt":null}
-  ]
-}

package/test-scenarios/v1.12.1-scenarios.md

@@ -1,42 +0,0 @@
-# v1.12.1 Provider Session Binding Scenarios
-
-Scenarios 146-158 are required. A real host session must not be simulated. When the host cannot return a canonical provider session ID, the expected result is blocked/not-verified with providerSessionId=null.
-
-## 146. Canonical provider session ID
-Verify that only a provider API returned canonical ID can enter providerSessionId.
-
-## 147. Ordinary thread ID rejection
-Use an ordinary Codex thread ID and verify it is rejected as provider identity.
-
-## 148. Project identity mismatch
-Use a session from another project root and verify bindingStatus=conflict or blocked.
-
-## 149. Role/task mismatch
-Use a session whose roleId or taskId differs and verify no binding is written.
-
-## 150. Title and cwd validation
-Verify required role title and project cwd before binding.
-
-## 151. Missing interaction evidence
-Verify that missing send or receive receipt keeps the record pending or blocked.
-
-## 152. Candidate reuse verification
-Verify old candidates are read and identity-checked before reuse; title alone is insufficient.
-
-## 153. Exclusive occupancy conflict
-Verify a session occupied by another project or role is blocked and audited.
-
-## 154. Manual binding without evidence
-Verify a user-confirmed manual entry without evidence remains pending/unknown and not executable.
-
-## 155. Recovery idempotency
-Verify recovery reuses taskId and roleTaskId, increments attempt, and creates no duplicate task or session.
-
-## 156. Writeback failure recovery
-Verify a created provider session with failed local writeback enters recovery-required and is queryable.
-
-## 157. Unsupported host degradation
-Verify unsupported host records local-degraded, providerSessionId=null, nextAction, and no fake ID.
-
-## 158. Cross-file binding consistency
-Verify provider binding, role task, assignments, task-index, and init-result agree on identity, status, attempt, and audit.