@abelxiaoxing/cadence 1.0.14 → 1.1.1

package/prompts/abel-implement.md

@@ -1,194 +1,138 @@
 ---
 name: abel-implement
-description: 校验已批准的 OpenSpec 变更，并调度子代理并行执行强制 Red-Green-Refactor，在任务检查点维护 AGENTS 索引。
+description: Execute one approved change through the durable control plane
 category: abel
-tags: [abel, implementation, TDD, agents]
+tags: [abel, implement, openspec, TDD, recovery]
 argument-hint: "<change_name>"
 ---
 
+Load the bundled `abel-workflow` Skill and read the complete value inside `<abel-request>` without tokenizing it a second time.
+
 <abel-request>
 $ARGUMENTS
 </abel-request>
 
 <!-- ABEL:PROMPT:abel-implement -->
 <!-- ABEL:START -->
-**参数**：从 `<abel-request>` 读取必需的 `<change_name>`。
-
-# 护栏
-
-- 改动保持最小且已获批准；应用任何修改前做副作用评审；保留无关的脏文件。
-- 最少注释/文档，优先自解释代码。
-- OpenSpec 约定以 `openspec/config.yaml` 声明的 schema 及其模板为准；OpenSpec 管理的 `openspec/` 生成物（schemas、`config.yaml`、skills 与 slash commands）绝不编辑。
-
-## AGENTS 阶段权限（Implement 专用）
-
-Design 阶段的 `AGENTS.md` 只读规则仅适用于 Design 阶段，不继承为全局规则。
-进入 Implement 后，父代理必须按 Gate B 已批准的结构化契约在稳定任务检查点执行：
-
-- `agentsImpact: none`：不得修改任何 `AGENTS.md`，且不得声明 `agentsTarget`。
-- `agentsImpact: update-existing`：父代理应更新指定现有索引的托管区块。
-- `agentsImpact: create-index`：父代理可创建批准路径的新索引。
-- `agentsImpact: remove-index`：父代理可移除批准路径的索引托管区块；仅当无人工内容残留时才删除文件。
-- 非 `none` 必须有明确 `agentsTarget` 且 `agentsManagedOnly: true`。
-  父代理只能修改批准路径及 `<!-- ABEL:AGENTS-INDEX:START -->` 到 `<!-- ABEL:AGENTS-INDEX:END -->` 的托管区域，保留全部人工内容。
-- 子代理始终不能修改任何 `AGENTS.md`，其派发写集必须排除 Gate receipt、OpenSpec 跟踪文件和 AGENTS 路径。
-- 已批准的 AGENTS checkpoint 是正常 Implement Green/检查点工作，必须完成；不能成为阶段路由或用户恢复动作。
-
-## TDD 护栏（强制）
-
-- **Red**：只创建/执行任务的失败可执行验证；实现代码禁止。
-- **Green**：写最小代码满足验证；过度工程禁止。
-- **Refactor**：在保持目标与受影响验证为绿、不引入完整套件新失败的前提下优化质量。
-- 每次代码/测试变更后必须运行任务验证，并在会话中重新分类累计的 AGENTS 影响；仅在稳定任务检查点写索引。
-  绝不跳过验证；测试先行强制。
-
-**技能集成**：见全局 AGENTS《阶段技能矩阵》Implement 列；E2E 任务 → `/dev-browser`。
-
-## 就绪预检（任何代码/测试写入之前）
-
-Implement 不重新作出或请求 Gate A/B 决策；只有 Gate receipts、artifact hashes、追溯链与 OpenSpec strict validation 全部通过时，才将所选变更视为可信设计交付。
-
-1. 从 `<abel-request>` 直接解析 change name；仅在缺失或无法唯一解析时提问。
-   绝不要求无条件的 `openspec view` 确认。
-2. 解析仓库根目录，记录 `git status --short`，阅读根/相关嵌套索引。
-   根索引缺失/无效 → `/abel-init`。
-3. `openspec status --change <name> --json`：读 `schemaName`、`changeRoot`、`artifactPaths`、`applyRequires`、`artifacts`。
-4. `openspec schema which <schemaName> --json` 读取解析出的 schema.yaml，提取 `apply.tracks` 并相对 `changeRoot` 解析；必须是 `changeRoot` 内一个已存在的具体常规文件。
-   缺失/空/不具体/越界路径一律关闭失败；绝不从 apply 指令的 `contextFiles`/`tasks` 推断跟踪路径。
-5. `applyRequires` 中每个产物 id 状态必须为 `done`；数组存在或总体完成标记不算数。
-6. 校验 `changeRoot/gate-a.yaml` 与 `changeRoot/ready.yaml`：当前 receipt 版本、change/schema 绑定、批准摘要、artifact 覆盖范围、`gate-a.yaml` hash，以及每个规范化相对路径的 SHA-256。
-   `ready.yaml` 必须只含一份 canonical `implementGraph`，且 `implementGraphHash` 必须等于 `hashImplementGraphBoundary(implementGraph)`；`verificationClosure` 必须精确为 `{ executable: true, diagnostics: [] }`。
-   hash 使用文件原始字节；仅对 schema `apply.tracks` 指向的任务文件先把 `- [x]`/`- [X]` 规范化为 `- [ ]`。
-   拒绝绝对路径、`..`、越界与符号链接逃逸。
-   receipt 缺失或任一 hash 失效即关闭失败。
-7. `openspec validate <name> --strict --type change` 问题为零。
-8. 阅读 `artifactPaths` 报告的全部规划产物；运行 `openspec instructions apply --change <name> --json` 并遵循其 apply 契约。
-9. 从 receipt 的 `implementGraph` 校验稳定 Requirement/Scenario 引用、Requirement → Scenario → Verification → Task 追溯链，以及每个任务的前置条件、`dependsOn`、计划写集、冲突/资源锁、派发上下文、验证、批准依赖、结构化 AGENTS 影响和影响闭包契约；缺失、含糊或不可判定时按下方矩阵关闭失败。
-   每个 phase 必须是 Gate B 批准的结构化 `kind: "vitest" | "package-script" | "static-check" | "steps"`；Implement 不从命令固定下标猜测语义，也不隐式运行 `bun run check`、`bun run test:target` 或任何 consumer script。
-   `steps` 按声明顺序执行；precheck 必须是显式 `expected-green` step，不得用 `&&` 或其他 shell operator 合成。
-   每个 phase 的直接 verification input 必须恰好绑定为 `workspace` path 或 graph `outputId`；每个 output 必须有唯一 id/path/producer phase 和 `regular-file` postcondition。
-   write set 只授予权限，不能推断 provenance。
-10. 影响闭包（impact closure）：若任务修改路由授权、页面状态、API 响应或公共 HTML，检查 URL/路由/handler/template 检索证据及全部相关既有测试归类。
-    E2E、theme/layout、authorization、HTML/template、API contract 测试必须进入当前写集、明确后续回归任务或有不受影响证据；affected suite 不能只包含新增测试。
-    `/videos`、`/api/videos` 等变化须检查全部这些测试面。
-11. 在 graph admission 前，使用 receipt 中同一个 `implementGraph` 重跑共享 `assessImplementGraphReadiness`。
-    fresh context 的 facts 只来自父代理在跟踪文件中拥有的 completed tasks、当前进程已知的 blocked tasks 与当前工作区安全扫描；不得因文件碰巧存在而推断 task/phase 完成。
-    静态 closure 必须继续为 `{ executable: true, diagnostics: [] }`；completed producer 的全部 outputs 必须实际为安全普通文件。
-    任一不一致都是 `delivery-invalid` stage blocker，不得调用方另算 missing inputs。
-    package script 的 package manager、名称与完整 command 必须和当前 consumer `package.json` 一致；本地 executable 必须受 `node_modules` 约束；`npx` 只能用 `noInstall: true` 并编译为 `--no-install`，不允许下载。
-    `vitest` 才注入 JSON reporter、应用 `minTests` 并解析 assertion Red identity；`package-script`/`static-check` 不附加 Vitest 参数，只使用退出码和批准的稳定输出 identity。
-    parent-only AGENTS checkpoint 可使用合法的 Node/static contract。
-    不支持的 shape 是 `design-readiness/verification-contract-unsupported` stage blocker；缺失 script 是 `verification-adapter/script-missing`，其他缺失/漂移能力使用精确 adapter code；只有 Bubblewrap、Bun 解析、依赖路径或 sandbox runtime 不可用才使用 environment code。
-12. 在任何写入前运行并记录每个未完成任务的目标命令、全部 affected-suite 命令与完整套件基线：命令、退出码、归一化失败标识/原因。
-    尚未加入 Red 验证时，目标命令可记录为契约规定的预期状态；既有失败与目标 Red 严格分离，既有失败绝不算 Red。
-13. 可信交付无效时，在注册任务前输出 stage-level blocker；不得伪装成 task outcome。
-    boundary 外的新路径、依赖、行为、策略、架构、冲突、资源、verification 或 AGENTS 需求使当前任务以 typed `approval-boundary` failure 终止。
-    工具、依赖路径、命令或沙箱不可用使当前任务以 typed environment failure 终止；可机械修复的产物仅使用 phase 的有界 correction budget。
-
-**工具路由**：父代理只负责可信交付校验、执行图调度、子代理派发、统一 diff 审查与机械应用、验证决策、Gate/AGENTS/任务状态所有权，不直接编写测试、实现或局部重构。
-任务的 Red、Green、Refactor 内容统一由任务 worker 子代理生成；E2E 任务使用 `/dev-browser`。
-
-## 执行图与并行调度（父代理专有）
-
-1. 以 receipt-bound `ImplementGraphBoundary` 为唯一执行图；校验 hash 后先调用一次 `admit-graph`，不得另建 stage graph 或逐任务重述完整 boundary。
-   从 graph tasks 的 `dependsOn` 计算 DAG，拒绝未知任务、自依赖和环。
-   非任务前置条件通过契约检查后，且所有前置任务均已由父代理验收并推进状态，任务才进入 ready 集合。
-2. 规范化每个计划写集、冲突集和资源锁；写集覆盖测试、fixture、快照与生成输出。
-   拒绝绝对路径、`..`、符号链接逃逸和不可判定的宽泛 glob。
-   Gate receipt、OpenSpec 跟踪文件和 `AGENTS.md` 不属于子代理写集，只能由父代理按契约写入。
-3. 父代理从 ready 集合计算受可用并发槽约束的批次；仅计划写集两两不相交、无冲突边且共享/验证资源锁兼容的任务可同批。
-   相同路径、祖先/后代目录、共同生成输出或互斥资源均串行；Design 的建议波次只供核对，不替代重新计算。
-4. 每个 ready task 的首次 Red `task-attempt` 由 Runtime 从已 admitted graph 打开一个 immutable TaskRecord；后续 Green、Refactor、artifact correction 或 stale refresh 仍只提交 `task-attempt` 的 change/task/request/phase identity 与 fresh snapshot。
-   每个 ready 任务绑定一个任务局部 worker，并在同一 worker 上通过后续派发连续完成 Red、Green、Refactor，避免把全局上下文复制给 worker。
-   首次派发只提供相关 AGENTS 内容、任务契约、允许读取/改动范围、可信基线标识与当前阶段；worker 不得广泛探索。
-5. 同批 worker 并行生成当前阶段的统一 diff 和说明，不得写入任何工作树。
-   父代理按跟踪文件顺序校验基线、实际路径、契约与 diff，再机械应用合格 diff；父代理不得自行补写或修正实现。
-   每次应用后由父代理运行契约命令，并把命令、退出码和归一化结果回传原 worker，作为进入下一阶段或生成修正 diff 的唯一事实。
-6. diff 冲突、越界、写集扩张或验证失败时只终止当前任务；已独立验收的同批任务可保留，外层 DAG 的其余调度仍由父代理决定。
-   机械重派只可在不改变行为、架构、依赖和固定 boundary 时基于最新 snapshot 进行。
-   普通实现或测试失败不得由自然语言推测为契约变化。
-   禁止无界重试。
-7. 一批任务稳定后，父代理逐任务完成 AGENTS 检查点和状态推进，再重新计算 ready 集合；任何后继不得提前调度。
-   循环直至 DAG 全部完成。
-8. 当前 phase candidate 在隔离 checkout apply 后、verification 前必须满足该 phase 的全部 output postconditions；main workspace apply 后、phase progression 前再次检查。
-   跨任务 output 仅在 producer 整个 task completed 后发布；producer blocked/failed 时 consumer 保持 `dependency-blocked` 且不启动 child，producer completed 但 output 缺失或 unsafe 时返回 `producer-output-unavailable`。
-
-## 每个任务的委派式 TDD 循环
-
-worker 消费任务契约的普通缩进符号列表（绝非 Markdown 复选框）：前置条件、直接依赖、派发上下文、计划写集、冲突/资源锁、验证类型、Red 命令/预期失败原因、Green 预期行为、受影响套件命令、目标范围和 AGENTS 影响。
-
-1. **🔴 Red（worker → 父代理）**：worker 只用契约规定的验证类型与范围生成失败验证 diff；父代理审查并应用后运行 Red 命令，必须因契约描述的目标缺陷失败。
-   加载/导入失败、未运行目标测试或失败身份与批准身份不同时，将结果分类为生成的实现产物拒绝并进入有界修正路径。
-   若 Red 候选意外通过，记录 `{ kind: "artifact", code: "red-not-witnessed" }`；Runtime 返回有界 `{ kind: "retry", scope: "worker", cause: "artifact", remainingAttempts: 1 }`，因为候选通过本身不证明设计有缺陷。
-   已证明批准命令无法见证行为且必须改变行为、策略、依赖、架构、范围、写集或验证契约时，以 `verification-contract-insufficient` 终止当前任务，不消耗产物修正预算。
-   非行为变更任务以指定的失败静态验证起步；仅人工任务不具备可执行合同并阻塞。
-2. **🟢 Green（worker → 父代理）**：同一 worker 基于已确认的 Red 证据生成最小实现 diff；父代理审查并应用后运行目标验证。
-   失败证据回传 worker，仅允许在声明写集和已批准行为内生成修正 diff。
-3. **🔵 Refactor（worker → 父代理）**：同一 worker 仅在声明范围内生成消除重复、改进命名/结构/可读性的可选 diff；父代理每次应用后跑目标验证，重构后跑 affected suite；失败则仅撤销该次重构 diff。
-4. **交付验收（父代理）**：拒绝基线不符、越界文件、Gate/索引/跟踪文件修改、验证证据缺失或未批准行为；父代理只做契约判断、机械应用与验证，不直接修码。
-5. **AGENTS 检查点（父代理）**：审查完整任务 diff，机械比较 `agentsImpact`、`agentsTarget`、`agentsManagedOnly: true` 与实际影响。
-   - `none` → 记录未写 AGENTS 的证据；`update-existing | create-index | remove-index` → 父代理在稳定任务检查点按批准目标完成托管区块操作。
-     子代理 diff 中出现 AGENTS 路径一律是写集越界。
-   - 实际 diff 暴露未批准生产行为/架构，或必须改变 AGENTS/任务契约时，以匹配的 `approval-boundary` code 终止当前任务；正常已批准 checkpoint 继续执行。
-   - 校验已索引的路径/命令/根到嵌套路由、marker 唯一性与人工内容不变；运行该任务的 AGENTS 验证命令。
-6. 父代理重跑目标与受影响验证；全绿后，仅在 schema `apply.tracks` 解析出的具体跟踪文件中更新匹配的那个任务复选框；零个/多个匹配 → stage-level blocker。
-
-## 当前任务的 typed blocker（闭合）
-
-Runtime 只报告当前任务事实，不选择用户恢复动作，也不控制外层 DAG。
-闭合失败类如下：
-
-- malformed diff、syntax/import/load、no-test/错误命令、wrong Red identity、重复/无效提交或 Red 候选意外通过：`{ kind: "artifact", code: <closed-artifact-code> }`，其中意外通过使用 `red-not-witnessed`；首次可在共享两次 launch 内修正，耗尽为 `attempts-exhausted`。
-- stale snapshot 使用 `{ kind: "stale", code: <closed-stale-code> }`；可机械重派的 transport failure 使用 `{ kind: "transport", code: "transport-failure" }`；首次可在共享两次 launch 内刷新或同调用重派，耗尽为 `attempts-exhausted`。
-- artifact、stale 或 transport 耗尽必须返回 `{ kind: "attempts-exhausted", cause, attemptsUsed: 2, lastFailure: { code, stage, details? } }`。
-  `lastFailure` 保留最终具体 closed code 与 `child-session-create | child-provider-stream | child-timeout | child-finalization | structural-submit | candidate-retention | candidate-diff | candidate-preflight | parent-review | candidate-apply | agents-checkpoint | phase-runtime` stage。
-  安全 details 只允许 final submit category、有界 submit attempts、schema state、request/role/task/phase 中不匹配的维度名和受约束 verification id；不得返回 prompt、diff、模型原始输出、excerpt、consumer 内容、command/argv、endpoint、credential、环境值或 identity 实际值。
-- Bubblewrap、依赖路径、测试沙箱或外部环境不可用：`{ kind: "environment", code: <closed-environment-code> }`，当前任务 terminal blocked。
-- 缺失/漂移的已批准 script、runner、本地 executable 或 verification input：`{ kind: "verification-adapter", code: <closed-adapter-code> }`；`script-missing` 不得误报为 `bubblewrap-or-dependency-unavailable`。
-- 新路径、依赖、行为、策略、架构、冲突、资源、verification 或 AGENTS 合同：对应 closed `approval-boundary` code，当前任务 terminal blocked，不扩大 boundary。
-- result size 超限：`{ kind: "result-limit", limitBytes }`，当前任务 terminal blocked，不接受 partial diff。
-
-可信 delivery 的 receipt/hash/trace/strict/graph closure failure 发生在 `admit-graph` 前并成为 `delivery-invalid` stage blocker，不是 TaskFailure。
-预期 Red 失败、artifact defect、stale snapshot、环境失败、已批准 AGENTS checkpoint、已批准文档/测试和 boundary 内兼容修复均不产生阶段选择。
-重试预算始终有限。
-
-父代理拥有主工作区补丁应用、闸门、索引写入与任务完成状态；子代理只接收任务局部索引上下文，只返回统一 diff 与分析，绝不应用补丁、批准决策、编辑索引或推进状态。
-
-## 最终评审与收尾
-
-1. 全部目标测试为绿后运行受影响套件。
-2. 重跑同一完整套件命令，与基线的归一化失败标识对比，要求无新增失败。
-3. 变更风险需要时以子代理做全局只读评审；接受其编辑后重跑目标验证与 AGENTS 分类。
-4. 目标/索引/受影响/完整套件任何失败都阻塞完成；只修复或回退本次变更的编辑。
-5. 最终副作用评审。
-6. 要求每个任务/索引结论均已解决，然后报告：Red/Green/Refactor 证据、受影响/完整套件相对基线的结果、AGENTS 文件与原因、是否可进入用户授权的归档。
-   绝不隐式归档。
-
-## 输出格式
-
-```text
-## /abel-implement (TDD)
-
-### 任务 i/N: {任务描述}
-🔴 Red  ├─ 类型: {verification_type} ├─ 运行: {red_command} └─ 结果: 因 {预期失败原因} 失败 ✓
-🟢 Green ├─ 实现: {文件} ├─ 运行: {red_command} └─ 结果: {green 预期行为} ✓
-🔵 Refactor ├─ 优化: {描述} ├─ 运行: {affected_suite_command} └─ 结果: 绿 ✓
-AGENTS: {none 的证据 | 更新的索引文件与原因}
-
-### 全部任务完成
-├─ 目标测试: 绿 ✓
-├─ 受影响套件: 绿 ✓
-├─ 完整套件: 相对基线无新增失败 ✓
-└─ 可进入用户授权的归档
+
+# Implement outcome
+
+The request must resolve to exactly one existing change.
+If it is missing, absent, or ambiguous, stop before execution and ask for the unique change name.
+
+Implement is a durable, resumable execution stage.
+The private control plane owns task scheduling, Red-Green-Refactor progression, candidate application, verification attribution, repair, task checkbox updates, AGENTS checkpoints, and the final transaction.
+The parent interprets state and makes product decisions; it must not reproduce those mechanics manually.
+
+## Trusted admission
+
+Before the first Worker launch, load the current `ready.yaml` and referenced canonical `implement-plan.json`; verify the Gate A binding, raw and canonical hashes, OpenSpec strict/planning status, Requirement → Scenario → Verification → Task traceability, capability closure, exact path/output contracts, repair policy, tracking contract, and sealed AGENTS operations.
+
+An invalid or incomplete delivery is one aggregated `delivery-invalid` admission result.
+Report every safe diagnostic together.
+Do not launch a Worker, consume a retry, mutate the main workspace, or infer a missing contract.
+
+Gate approval is not tool permission.
+A fresh context validates existing approval facts and does not ask the user to approve unchanged decisions again.
+
+## Closed control surface
+
+Use `abel_dispatch` only with the change commands below; never send `version`; never send `action: "run"`; and never send `admit-graph`, `task-attempt`, caller-owned snapshots, completed/blocked arrays, or apply identities.
+
+```json
+{"command":"start","stage":"abel-implement","change":"<change>","operationId":"<unique-operation>"}
+{"command":"status","stage":"abel-implement","change":"<change>"}
+{"command":"resume","stage":"abel-implement","change":"<change>","operationId":"<unique-operation>"}
+{"command":"rebind","stage":"abel-implement","change":"<change>","operationId":"<unique-operation>","routeId":"<approved-route>"}
+{"command":"cancel","stage":"abel-implement","change":"<change>","operationId":"<unique-operation>"}
+{"command":"discard","stage":"abel-implement","change":"<change>","operationId":"<unique-operation>"}
 ```
 
-<!-- ABEL:END -->
+When the user approves a revised delivery boundary, resume the same run with both `deliveryRevision` and `receiptHash`.
+The engine revalidates retained phase/repair facts, invalidates only incompatible tasks, rebuilds tracking, and preserves compatible work.
+
+`status` is local and must remain useful when every Worker endpoint is unavailable.
+Repeating the same operation id returns its committed outcome.
+Use a new operation id for a new attempt.
+
+For an approval-needed retained run, a later explicit Implement invocation must inspect local `status`.
+When a newer owner-private-proof-bound receipt is present, status exposes its exact pair in `availableDelivery`, adds `resume` to `legalCommands`, and fills the matching `conditionalCommands` arguments.
+Resume with that discovered pair; do not require the user to copy revision or hash data from a prior Design conversation.
+Discovery never admits the delivery: `resume` still performs the full artifact, traceability, capability, and currentness validation before any Worker starts.
+
+## Execution and verification
+
+Before any candidate, the engine records target-contract, task-affected, and full-suite baselines with normalized failure identities.
+A pre-existing failure remains separate and never satisfies Red or becomes evidence that this change caused it.
+
+For each ready task the engine runs:
+
+1. Red in a disposable child revision; the declared target must fail with the approved identity.
+2. Green with the minimum approved implementation; the declared target and task-affected contract must pass relative to baseline.
+3. Optional Refactor with behavior unchanged and verification still green.
+4. Parent-owned task completion and exactly one `tasks.md` checkbox update in the private cumulative revision.
 
-If the required input is missing or absent, or the request is ambiguous and not unique, stop before any work and ask the user for the missing or clarified input.
+Independent tasks may merge only after currentness and declared conflict checks.
+Conflicting tasks remain durably queued without spending a Worker attempt.
+Workers only propose complete diffs inside their exact write/delete sets; they never write the workspace, AGENTS, task state, or the main repository.
 
-Record target, affected suite, and full-suite baselines before writing; keep pre-existing failures separate and never attribute them to the task Red.
-Red must fail with the expected identity; a wrong reason is a generated implementation-artifact rejection handled by bounded correction.
-If a Red candidate passes, classify it as an artifact failure with `code: red-not-witnessed`; a separately proven insufficient verification contract terminally blocks the current task without consuming the artifact correction budget.
-After Green the target and affected suite must be green, in Red-Green-Refactor order, with a stable AGENTS index at checkpoints and no new failure relative to the recorded baseline.
-A fresh-context handoff validates the Gate receipt, hash, and trace strictly without requesting Gate approval again.
-You must not archive, publish, or commit implicitly: only explicit parent actions may do so.
+After all tasks, run every task-affected contract, the approved full suite, output postconditions, and the code-owned managed-only AGENTS checkpoint.
+Only then prepare one journaled, currentness-checked cumulative apply and the declared post-apply verification.
+The main workspace stays unchanged before that transaction.
+
+## Failure attribution and recovery
+
+Ordinary failures stay inside this Implement run:
+
+- `artifact`, wrong-Red identity, malformed/incomplete diff, stale candidate, or transport failure: use the bounded automatic attempt policy; after exhaustion pause with the last safe code and retain the last committed revision.
+- Artifact correction alone consumes the sealed `artifactCorrection.maxAttempts` of 2-3 total candidate launches per task phase and operation, including the initial launch; a later `resume` starts a fresh operation budget.
+- `environment` or `verification-adapter`: pause without speculative edits.
+  Resume after the capability is restored; baseline facts and completed phases are reused.
+- `pre-existing`: keep as baseline evidence.
+  It does not authorize a code change and does not fail completion unless the change worsens it.
+- `introduced` in a task-affected contract: reopen the owning task as `repairable`, perform the minimum bounded repair inside its approved paths, record `repair-verified`, and re-run attribution automatically.
+- `unresolved` full-suite attribution: pause with failure identities and retained cumulative state; resume after evidence resolves ownership.
+- `workspace-revision-stale`, conflict, cancellation, process interruption, or apply recovery: preserve user edits and resume from the last durable checkpoint.
+  Cancellation is budget-neutral.
+  During a partially visible final apply, recovery must settle before pause or discard.
+- `result-limit`: pause as `needs-task-split`; never accept a partial diff.
+
+No item above carries stage-routing metadata or asks the user to restart the approved workflow.
+
+## The only approval-needed boundary
+
+Use `approval-needed` only when continuing requires authority not present in the sealed delivery: new observable behavior, architecture or policy; a new/changed dependency; an undeclared write/delete/read target; expanded conflict/resource authority; a changed verification contract; an AGENTS target/impact/managed block change; or another irreversible scope decision.
+
+Classify only codes in the closed authority table.
+An unknown Worker, verifier, or internal approval code is an `approval-code-invalid` integrity pause; never guess a category or Gate from keywords.
+
+Report the exact missing authority and retained run state.
+Do not discard compatible work.
+Expose `approval.category`, `approval.requiredGates`, safe `approval.refs`, and `approval.designRequest` as `/abel-design --change <change>` user guidance only; never invoke Design automatically.
+Bare `resume` is not immediately executable progress while approval is missing, so keep it out of `legalCommands` and expose it only in `conditionalCommands` with a `deliveryRevision` greater than the retained revision and the matching `receiptHash` precondition.
+If the user supplies that newly approved receipt revision, resume the same run with the revision/hash pair.
+A generated artifact defect, endpoint outage, stale snapshot, environment failure, pre-existing failure, introduced in-boundary repair, approved documentation/test edit, or approved AGENTS checkpoint is never `approval-needed`.
+
+## AGENTS and tracking
+
+Design-time AGENTS read-only authority does not carry into Implement.
+Gate B seals `none | update-existing | create-index | remove-index`, exact targets, owning task ids, verification, and the complete managed block.
+The parent control plane applies those operations only in the private cumulative revision, preserves all human text outside `<!-- ABEL:AGENTS-INDEX:START -->` and `<!-- ABEL:AGENTS-INDEX:END -->`, and never gives an AGENTS write path to a Worker.
+
+Task checkboxes are parent-owned progress facts.
+The loader may normalize only `[x]`/`[X]` to `[ ]` for the sealed task ids when checking the delivery hash; every other byte remains integrity-bound.
+
+## Truthful completion
+
+Treat `queued`, `connecting`, `waiting-first-response`, `running`, `validating`, `retrying`, `verifying`, `paused`, `approval-needed`, `applying`, and `recovering` as nonterminal activity states.
+`operation-cancelled` ends only the current operation; `discarded` and `rejected` are non-success terminal run states.
+Tool-call settlement is not success.
+Report success only when the durable state is `completed`, final application and post-apply verification committed, and private run content was cleaned.
+Terminal `completed`, `discarded`, or `rejected` settlement deactivates private dispatch; a nonterminal pause or approval wait keeps it active for direct follow-up.
+
+On a nonterminal result, report the safe code, affected task/scope when present, retained progress, and the exact legal command that continues or inspects the same run.
+Never hide an incomplete run behind a success checkmark.
+
+Do not archive, publish, release, stage, or commit implicitly.
+
+<!-- ABEL:END -->

package/prompts/abel-init.md

@@ -1,42 +1,45 @@
 ---
-description: Initialize OpenSpec and AGENTS indexes for an Abel workflow project
+description: Initialize or safely repair an Abel OpenSpec project
 argument-hint: "[project-path]"
 ---
 
-Load the bundled `abel-workflow` Skill before inspecting or modifying OpenSpec or AGENTS files.
-If that core Skill is unavailable, stop before any write and report an executable package-resource restore or reinstall command.
-Read the complete value inside `<abel-request>` without a second whitespace-tokenization pass.
+Load the bundled `abel-workflow` Skill before inspecting or writing the target.
+If it is unavailable, stop before writes and report one executable package restore or reinstall command.
+Read the complete value inside `<abel-request>` without tokenizing it a second time.
 
 <abel-request>
 $ARGUMENTS
 </abel-request>
 
-The request is an optional project path; when absent, target the current working directory.
-Resolve that root before action and preserve unrelated files, all human-authored AGENTS content, baseline dirty state, and nested repository boundaries.
-Never overwrite or cross a nested repository boundary merely because it is below the selected root.
-
-## Prerequisites
-
-Select Bun when it is usable; otherwise select npm.
-Use that single JavaScript toolchain for the whole run.
-If neither is usable, stop with the original probe errors and executable remediation.
-
-Discover the bundled `context7-auto-research` and `grok-search` Skills.
-Record each resolved discovery path when present.
-When either is missing, report an actionable package-resource restore or reinstall command, continue OpenSpec initialization and AGENTS repair, and mark the final environment as not fully ready.
-A missing research Skill does not prevent OpenSpec initialization or AGENTS repair.
-The absence of `git-commit`, external `dev-browser`, or a dedicated `time` Skill does not affect Init readiness; do not check them as Init prerequisites.
-
-Probe OpenSpec and all CLI capabilities required by the shared workflow.
-If OpenSpec is absent, install `@fission-ai/openspec@latest` globally with the selected toolchain (`bun add --global @fission-ai/openspec@latest` or `npm install --global @fission-ai/openspec@latest`), then recheck every required capability.
-If installation or the recheck fails, stop, preserve the original error, and report the exact executable remediation command for the selected toolchain.
-
-## Initialization
-
-Initialize or safely update OpenSpec without `--force` or another destructive force option.
-Resolve and validate the selected schema and its templates before reporting success.
-Create or repair only the verified managed blocks of appropriate root and nested AGENTS indexes, preserving all human-authored text and unrelated content.
-You must not edit `openspec/AGENTS.md`.
-
-Report the resolved project root, selected toolchain, OpenSpec capability results, schema/template validation, AGENTS repairs, research-Skill discovery paths or remediation, and final readiness.
-A missing research Skill makes the environment not fully ready even when OpenSpec and AGENTS repair completed.
+The request is an optional project path; when absent, use the current working directory.
+Resolve and report the canonical target root before action.
+
+## Deterministic boundary
+
+- Init performs no Subagent or `abel_dispatch` work.
+  Probe and repair locally in a fixed order.
+- Preserve the baseline dirty state, unrelated files, all human-authored AGENTS text, and every nested repository boundary.
+- Never use `--force`, replace an existing OpenSpec tree wholesale, follow a symlink escape, or edit `openspec/AGENTS.md`.
+- Choose Bun when usable, otherwise npm, and use only that toolchain for this run.
+  If neither works, stop with the original probe errors and an executable remediation.
+
+## Fixed procedure
+
+1. Inspect the canonical root, repository boundaries, OpenSpec presence, project configuration, and root/nested AGENTS marker structure without writing.
+2. Probe the required OpenSpec CLI capabilities, schema selection, schema validation, templates, status, instructions, and strict validation.
+3. If OpenSpec is absent, install `@fission-ai/openspec@latest` globally with the selected toolchain (`bun add --global @fission-ai/openspec@latest` or `npm install --global @fission-ai/openspec@latest`), then repeat the complete capability probe.
+   If installation or recheck fails, preserve the original error and stop with the exact remediation command.
+4. Initialize only when no OpenSpec root exists.
+   Otherwise repair only missing or mechanically invalid configuration that can be changed without discarding project choices.
+5. Resolve and validate the selected schema and templates before claiming readiness.
+6. Create or repair only verified `<!-- ABEL:AGENTS-INDEX:START -->` … `<!-- ABEL:AGENTS-INDEX:END -->` managed regions.
+   Preserve every byte of human content outside those markers and do not cross nested repositories.
+7. Re-run all probes after writes.
+   A second identical Init must produce no additional changes.
+
+Discover the bundled `context7-auto-research` and `grok-search` Skills and report their resolved paths.
+Their absence does not prevent OpenSpec or AGENTS repair, but final readiness is `partial` and must include an actionable package-resource restore command.
+Do not probe `git-commit`, external `dev-browser`, or a dedicated time Skill; none is an Init prerequisite.
+
+Report one final result with the root, selected toolchain, actions actually taken, OpenSpec capability/schema/template evidence, AGENTS files changed or unchanged, research-Skill paths or remediation, and `ready | partial | paused`.
+Never report success from the initial probe when the post-write recheck did not pass.