@c4a/context-cli 0.6.18 → 0.7.0

package/README.md

@@ -21,9 +21,10 @@ npm install -g @c4a/context-cli@latest
 context plugin install
 ```
 
-Restart or refresh the Agent host after installation. The bundled installer
-projects the same source entry into supported Claude, Codex, Cursor, and
-Skill-compatible layouts.
+Restart or refresh the Agent host after installation. One install creates the
+host-namespaced Context entry and projects `context-code-indexer` and
+`context-markdown-indexer` as unnamespaced lifecycle Provider Skills for
+Claude, Codex, and Cursor.
 
 The public community entry is `/c4a:context`. It creates a requested workspace,
 locates an existing workspace, or resumes the current workflow. Users should
@@ -169,11 +170,74 @@ for maintainers, automation, and diagnostics:
 - `context plugin status` checks installed Agent projections;
 - `context debug enable` records optional traces below
   `.tmp/context-runtime/debug/`;
-- new workspaces enable conservative document revisions by default;
+- new workspaces enable source-constrained editorial revisions by default;
   `context optimize-docs enable|disable` changes that workspace preference
   without modifying approved `knowledge/`;
 - `context revise "<title or approved path>"` starts one conversational
   correction and routes its validation before the next package build;
+- code extraction writes `knowledge/codeindex/**` and independently audits
+  source analysis, stable-boundary coverage, content density, evidence scope,
+  and page size before Review; legacy `codegraph` workspaces migrate through
+  the returned Route action;
+- the `context indexer` lifecycle builds contract-derived question targets,
+  digest-bound partition and author worksets, exact SubjectKey resolution
+  views, and independent post-author composer worksets. Main runs use a local
+  content-addressed ledger whose accepted Result/receipt transition is atomic;
+  ordinal/fixed-count partitions advance through the authorized strategy order,
+  and exhausted strategies use a persisted-predecessor CLI catalog fallback that
+  creates one parent unit without an Agent or user Gate;
+  post-author runs use a separate ledger and atomic envelope record. Observe
+  commands publish pending/accepted/failed/stale counts and never use numeric
+  cursors. Reconciliation then recomputes every required coverage domain from
+  current owner cells, accepted author Results, registered materials, and the
+  complete CLI-owned question-target set; capability or material gaps cannot be
+  reported as complete;
+- the `markdown-provider` Indexer Route starts only from exact current document
+  capture. Agent discovery reports visible `context-markdown-indexer*` Skills
+  without persisting the discovery list; the CLI recomputes the route and
+  static validation, resolves and stages exact CLI-bundled Bundles, and stops
+  external resolution or local customization at explicit Host/customization
+  Outcomes;
+- `route-index-requirement-confirmation` recomputes the canonical requirement
+  comparison before routing. Ordinary confirmation may use its explicit
+  authority, while contraction or incomparable obligation replacement always
+  enters the human-only Gate. `validate-subject-key-schemas` likewise requires
+  an exact Provider-major re-identification authorization for identity-breaking
+  changes over approved Nodes; incomplete, split, merged, or colliding mappings
+  fail before that Gate. Exact target matches that resolve to multiple Nodes
+  return `index-target-resolution-ambiguous`, and invalid reuse returns
+  `index-target-resolution-invalid`;
+- `context indexer build-material-question-workset` derives second-phase answer
+  eligibility from current registry, Provider, scope, source, and evidence-kind
+  authority. `prepare|observe|start|accept|fail-material-answer-*` use an
+  independent durable ledger; complete empty Results are cache hits, while
+  candidate evidence requires exact current source-span read receipts;
+- material-answer candidates use a separate limited Review Route. Its baseline
+  and decision bind one exact question target and canonical source-span evidence
+  set. Approval returns an `answer-approved` successor-ledger fact for the later
+  checkpoint writer; it does not approve an Artifact, reader page, or final
+  knowledge candidate;
+- main Candidate Review has one Graph predecessor:
+  `inspect-index-candidate-review-readiness`. The CLI accepts only digests of
+  content-addressed precompile/postcompile audit records and verifies their
+  requirement, registry, inventory, layout, candidate-set, and effective
+  revision bindings. Missing, stale, baseline-failed, or profile-blocked audits
+  cannot resolve the Review Gate. A profile failure enters
+  `revise-index-output`; `record-index-profile-revision` retains a stable
+  problem lineage and at most three distinct result fingerprints. After the
+  third failure, `report-index-profile-failure` persists the full metrics,
+  examples, history, missing inputs, and capability-loss report.
+  `inspect-index-profile-failure` is read-only, and the following
+  `override-index-profile-audit` Gate is non-delegable. Its receipt binds the
+  report, audit, candidate revision, failed metrics, user, and timestamp.
+  Baseline failures cannot issue or consume that receipt;
+- `checkpoint-material-gaps` is the mandatory retained writer before any
+  reconciliation stop. Approved answers inline their accepted workset;
+  `actualize-material-answer-bindings` maps them against one current layout
+  proposal set and reopens stale sources, while `close-indexer-approved-knowledge`
+  accepts provenance only from the approved structure it closes atomically.
+  `audit-material-gap-state` is read-only: it recomputes the expected ledger and
+  routes pre-Review drift to checkpoint or approved post-Review state to close;
 - `context clean-cache --dry-run` previews cleanup of Context-owned stale
   plugin caches.
 
@@ -181,11 +245,11 @@ Revision-bound commands should be copied from `workflow.current`; examples in
 documentation are orientation, not a substitute for the current Route. See the
 [installed quickstart](./docs/quickstart.md) and
 [debug tracing guide](./docs/debug-tracing.md).
-See also [document revisions](./docs/document-optimization.md).
+See also [source-constrained editorial revisions](./docs/document-optimization.md).
 
 ## Documentation and development
 
-- [Plugin contract](./plugin/README.md)
+- [Plugin contract](../../plugins/context/README.md)
 - [Workflow Provider internals](./context-workflow/README.md)
 - [SDK documentation index](../context/docs/README.md)
 - [Knowledge-project walkthrough](../context/docs/getting-started.md)
@@ -203,8 +267,9 @@ bun run --filter @c4a/context-cli lint
 bun run --filter @c4a/context-cli test
 ```
 
-Build generates the installable host projections under `dist/plugins`; edit
-only `plugin/` and the bundled workflow source, never generated output.
+Build generates installable host projections under `dist/plugins` and the
+direct-Git projection under `../../plugins/context/repo-install`; edit only
+`../../plugins/context/` and the bundled workflow source, never generated output.
 
 ## License
 

package/README.zh-CN.md

@@ -17,8 +17,9 @@ npm install -g @c4a/context-cli@latest
 context plugin install
 ```
 
-安装后重启或刷新 Agent 宿主。同一份入口源会被安装器投影到支持的 Claude、Codex、
-Cursor 和 Skill-compatible 目录。
+安装后重启或刷新 Agent 宿主。一次安装会生成一个带宿主命名空间的 Context 主入口，
+并把 `context-code-indexer`、`context-markdown-indexer` 作为无命名空间的生命周期
+Provider Skill 投影到 Claude、Codex 和 Cursor 的用户技能目录。
 
 社区版公开入口是 `/c4a:context`。它可以创建用户请求的工作区、定位已有工作区，
 或从当前状态继续流程。用户应该从知识意图开始，而不是内部命令：
@@ -143,20 +144,58 @@ context/
 - `context <command> --help` 是当前参数的权威来源；
 - `context plugin status` 检查已安装的 Agent 投影；
 - `context debug enable` 在 `.tmp/context-runtime/debug/` 记录可选轨迹；
-- 新工作区默认开启保守的文档修订；`context optimize-docs enable|disable`
+- 新工作区默认开启来源约束的编辑修订；`context optimize-docs enable|disable`
   可调整工作区偏好，同时保持 `knowledge/` 中的正式知识不变；
 - `context revise "<标题或正式知识路径>"` 从对话启动一次单页修正，校验后再由
   Route 提示重新构建；
+- 代码提取写入 `knowledge/codeindex/**`，并在 Review 前分别检查输入分析、稳定边界
+  覆盖、正文密度、证据范围和页面规模；旧 `codegraph` 工作区只通过 Route 返回的
+  迁移动作升级；
+- `context indexer` 生命周期以内容寻址 ledger 恢复 main/post-author 运行，再由 CLI
+  根据 current owner、accepted Result 和完整 question-target 集重新计算 coverage；
+  capability/material gap 不能被报告为 complete。ordinal/fixed-count partition 按授权顺序自动重试；
+  策略耗尽后只允许带持久化前驱的 CLI catalog fallback 生成唯一父单元，不调用 Agent 或用户 Gate；
+- `markdown-provider` Indexer Route 只能从精确 current 的文档 capture 开始。Agent
+  只报告当前会话可见的 `context-markdown-indexer*` Skill，不持久化 discovery list；
+  CLI 重新计算 Route 与静态校验，自动解析并 stage 精确 CLI-bundled Bundle，外部解析
+  或本地定制则分别停在显式 Host/customization Outcome；
+- `route-index-requirement-confirmation` 会重新计算 canonical requirement comparator；普通确认
+  只可使用显式 authority，contraction 或不可比较的义务替换始终进入不可委托人工 Gate。
+  `validate-subject-key-schemas` 对已有 approved Node 上的 identity-breaking 变化要求 Provider
+  major 与精确 re-identification 授权，漏映射、split、merge 或 collision 在 Gate 前失败；精确
+  target 命中多个 Node 返回 `index-target-resolution-ambiguous`，非法复用返回
+  `index-target-resolution-invalid`；
+- `context indexer build-material-question-workset` 从 current registry、Provider、scope、
+  source 与 evidence-kind authority 的交集派生二阶段回答资格；
+  `prepare|observe|start|accept|fail-material-answer-*` 使用独立 durable ledger，完整空 Result
+  也作为 cache hit，candidate evidence 则必须携带精确的 current source-span 读取回执；
+- material-answer candidate 使用独立的受限 Review Route，只批准一个精确问题目标与
+  canonical source-span evidence binding。批准结果只是供后续 checkpoint 写入的
+  `answer-approved` successor-ledger fact，不批准 Artifact、读者页面或主知识候选；
+- 主 Candidate Review 只有 `inspect-index-candidate-review-readiness` 一个 Graph 前驱。CLI
+  只接受内容寻址 precompile/postcompile audit record 的 digest，并复核 requirement、registry、
+  inventory、layout、candidate set 与 effective revision 绑定；record 缺失、stale、基础失败或
+  profile 未通过时都不能解析 Review Gate。profile 失败进入 `revise-index-output`；
+  `record-index-profile-revision` 按稳定 problem lineage 最多记录三个不同的 Result fingerprint。
+  第三轮后 `report-index-profile-failure` 才持久化包含指标、示例、历史、缺失输入与能力损失的完整
+  报告；`inspect-index-profile-failure` 只读展示该报告，随后
+  `override-index-profile-audit` 使用不可委托 Gate。回执精确绑定报告、审计、候选 revision、失败
+  指标、用户和时间；baseline 失败不能生成或消费该回执；
+- `checkpoint-material-gaps` 是 reconciliation 任何停止前的 retained writer。批准答案会
+  内联 accepted workset；`actualize-material-answer-bindings` 按同一个 current layout proposal
+  set 映射并在 source stale 时 reopen，`close-indexer-approved-knowledge` 只接受其同时关闭的
+  approved structure 内部 provenance，并原子收敛 ledger；`audit-material-gap-state` 始终只读，
+  内存复算 expected ledger，并把主 Review 前的漂移路由到 checkpoint、把已批准的后置状态路由到 close；
 - `context clean-cache --dry-run` 预览 Context-owned 过期插件缓存清理。
 
 绑定 revision 的命令应从 `workflow.current` 原样复制；文档示例只用于理解入口，
 不能替代当前 Route。参见[安装后快速说明](./docs/quickstart.zh-CN.md)和
 [debug tracing](./docs/debug-tracing.md)。
-文档编译优化与对话修订参见[文档修订页](./docs/document-optimization.zh-CN.md)。
+文档编译优化与对话修订参见[来源约束的编辑修订](./docs/document-optimization.zh-CN.md)。
 
 ## 文档与开发
 
-- [插件契约](./plugin/README_CN.md)
+- [插件契约](../../plugins/context/README_CN.md)
 - [Workflow Provider 内部说明](./context-workflow/README.zh-CN.md)
 - [SDK 文档索引](../context/docs/README.zh-CN.md)
 - [知识项目完整示例](../context/docs/getting-started.md)
@@ -173,8 +212,9 @@ bun run --filter @c4a/context-cli lint
 bun run --filter @c4a/context-cli test
 ```
 
-构建会在 `dist/plugins` 生成可安装的宿主投影；只修改 `plugin/` 和 Workflow
-Provider 源码，不要直接编辑生成产物。
+构建会在 `dist/plugins` 生成 npm 宿主投影，并在
+`../../plugins/context/repo-install` 生成 Git 直装投影；只修改
+`../../plugins/context/` 和 Workflow Provider 源码，不要直接编辑生成产物。
 
 ## License