feat(context): rebrand the context file CLAUDE.md → CLAWCODEX.md

FEATURE_LIST.md

@@ -41,7 +41,7 @@
 | Session message management | ✅ | Supports session history maintenance and serialization |
 | Error recovery / re-login | 🟡 | Basic auth error handling and reconfiguration flow exists |
 | Token / Cost tracking | 🚫 | Chat CLI does not yet have a complete statistics view |
-| Context building | 🟡 | Basic `context_system` supports workspace / git / `CLAUDE.md` injection; still missing README summary, memory, compact |
+| Context building | 🟡 | Basic `context_system` supports workspace / git / `CLAWCODEX.md` injection; still missing README summary, memory, compact |
 | Claude Code Agent Loop | ✅ | Implemented in agent_loop.py, supports tool call loops |
 | `/resume` session recovery | 🚫 | No standalone recovery flow or UI yet |
 | `/compact` conversation compaction | 🚫 | No automatic/manual compaction capability yet |
@@ -114,7 +114,7 @@
 | Tool Execution Engine | ✅ | Tool loading, execution, and result feedback loop implemented |
 | Output Styles | ✅ | Output style loading system implemented |
 | Session Persistence | ✅ | Session save/load capability exists |
-| Context Engine | 🟡 | Basic context building pipeline connected, supports workspace, git, `CLAUDE.md` prompt injection |
+| Context Engine | 🟡 | Basic context building pipeline connected, supports workspace, git, `CLAWCODEX.md` prompt injection |
 | Permission Engine | 🟡 | Framework exists, not fully integrated into tool execution flow |
 | Compaction Engine | 🚫 | No conversation compaction or token management capability |
 | Hook Runtime | 🚫 | No settings-driven hook execution mechanism |
@@ -201,7 +201,7 @@ Goal: Move from "mirrored tool catalog" to "genuinely executable Python Agent".
 Goal: Complete Claude Code's engineering capabilities.
 
 - [ ] Workspace context building improvements
-- [x] git status / file tree / `CLAUDE.md` injection (basic version)
+- [x] git status / file tree / `CLAWCODEX.md` injection (basic version)
 - [ ] README / entry file summary injection
 - [ ] Memory and historical context management
 - [ ] Full permission system integration
@@ -351,7 +351,7 @@ You can introduce the project like this:
 | 会话消息管理 | ✅ | 支持会话历史维护与序列化 |
 | 错误恢复 / 重新登录 | 🟡 | 已有基础认证错误处理与重新配置流程 |
 | Token / Cost 跟踪 | 🚫 | 当前聊天 CLI 尚未形成完整统计视图 |
-| 上下文构建 | 🟡 | 已有 `context_system` 基础版，支持 workspace / git / `CLAUDE.md` 注入，仍缺 README 摘要、memory、compact |
+| 上下文构建 | 🟡 | 已有 `context_system` 基础版，支持 workspace / git / `CLAWCODEX.md` 注入，仍缺 README 摘要、memory、compact |
 | Claude Code Agent Loop | ✅ | 已实现 agent_loop.py，支持工具调用循环 |
 | `/resume` 会话恢复体验 | 🚫 | 暂无独立恢复流程与 UI |
 | `/compact` 对话压缩 | 🚫 | 暂无自动/手动压缩能力 |
@@ -424,7 +424,7 @@ You can introduce the project like this:
 | Tool Execution Engine | ✅ | 已实现工具加载、执行、结果回填闭环 |
 | Output Styles | ✅ | 已实现输出样式加载系统 |
 | Session Persistence | ✅ | 已有会话保存/加载能力 |
-| Context Engine | 🟡 | 已接入基础上下文构建链路，支持 workspace、git、`CLAUDE.md` prompt 注入 |
+| Context Engine | 🟡 | 已接入基础上下文构建链路，支持 workspace、git、`CLAWCODEX.md` prompt 注入 |
 | Permission Engine | 🟡 | 已有框架，未完全集成到工具执行流程 |
 | Compaction Engine | 🚫 | 未形成对话压缩与 token 管理能力 |
 | Hook Runtime | 🚫 | 未接入设置驱动的 hook 执行机制 |
@@ -511,7 +511,7 @@ You can introduce the project like this:
 目标：补齐 Claude Code 的工程化能力。
 
 - [ ] 工作区上下文构建完善
-- [x] git status / 文件树 / `CLAUDE.md` 注入基础版
+- [x] git status / 文件树 / `CLAWCODEX.md` 注入基础版
 - [ ] README / 入口文件摘要注入
 - [ ] memory 与历史上下文管理
 - [ ] 权限系统完全集成

README.md

@@ -457,7 +457,7 @@ Use `clawcodex config` to check connection status and
 | Skill System | ✅ | SKILL.md-based slash-command skills with args + tool limits |
 | Cancellation / Abort | ✅ | ESC closes in-flight Bash, Grep/Glob, and streaming HTTP within ~50ms across every provider; subagents get isolated `AbortController`s; `Bash` `tool_result` distinguishes timeout from ESC-abort |
 | Image Handling | ✅ | TS-parity Read pipeline (magic-byte sniff, resize/compress to API limits); `@image.png` @-mentions inline as `ImageBlock`; pre-API base64 size validation in `BaseProvider._prepare_messages`; binary @-mentions (PDF/zip/docx/...) routed to a Read-tool hint instead of mojibake |
-| Context Building | 🟡 | Workspace / git / `CLAUDE.md` injection; richer summaries and memory still evolving |
+| Context Building | 🟡 | Workspace / git / `CLAWCODEX.md` injection; richer summaries and memory still evolving |
 | Permission System | 🟡 | Framework and checks; full integration still in progress |
 | MCP | 🟡 | MCP-oriented tools and wiring; full protocol/runtime polish ongoing |
 
@@ -782,7 +782,7 @@ clawcodex/
 │   ├── tool_system/        # Agent loop, tools, schemas
 │   ├── skills/             # SKILL.md loading and skill tool
 │   ├── services/           # MCP, compact, IDE bridge, tool execution, …
-│   ├── context_system/     # Workspace / git / CLAUDE.md context
+│   ├── context_system/     # Workspace / git / CLAWCODEX.md context
 │   ├── permissions/        # Permission modes and bash parsing
 │   ├── hooks/              # Hook types and execution helpers
 │   └── command_system/     # Slash commands and substitution

docs/ARCHITECTURE.md

@@ -110,7 +110,7 @@ subsumed under the bash tool path.
 
 Three tiers, matching the book chapter:
 
-1. **Project-level** — `CLAUDE.md` files in the repo (loaded by
+1. **Project-level** — `CLAWCODEX.md` files in the repo (loaded by
    `src/memdir/memory_scan.py`).
 2. **User-level** — `~/.clawcodex/projects/<project-slug>/memory/MEMORY.md` (loaded by
    `src/memdir/paths.py`).

docs/i18n/README_ZH.md

@@ -326,7 +326,7 @@ clawcodex --allow-dangerously-skip-permissions         # 允许之后通过 /per
 | Skill 系统 | ✅ | 基于 SKILL.md 的斜杠技能：参数与工具白名单 |
 | 取消 / 中止 | ✅ | ESC 可在约 50ms 内中止进行中的 Bash、Grep/Glob 以及所有 provider 的流式 HTTP；子 agent 拥有隔离的 `AbortController`；`Bash` 的 `tool_result` 区分超时与 ESC 中止 |
 | 图像处理 | ✅ | 与 TS 对齐的 Read 管线（魔数嗅探、按 API 限制缩放/压缩）；`@image.png` @-提及内联为 `ImageBlock`；`BaseProvider._prepare_messages` 中调用 API 前的 base64 大小校验；二进制 @-提及（PDF/zip/docx/…）转为 Read 工具提示而非乱码 |
-| 上下文构建 | 🟡 | workspace / git / `CLAUDE.md` 注入；更丰富的摘要与 memory 仍在演进 |
+| 上下文构建 | 🟡 | workspace / git / `CLAWCODEX.md` 注入；更丰富的摘要与 memory 仍在演进 |
 | 权限系统 | 🟡 | 框架与检查逻辑已有；全面集成进行中 |
 | MCP | 🟡 | MCP 相关工具与接线已有；协议层/运行时仍在完善 |
 
@@ -621,7 +621,7 @@ clawcodex/
 │   ├── tool_system/        # Agent loop、工具与 schema
 │   ├── skills/             # SKILL.md 加载与 Skill 工具
 │   ├── services/           # MCP、compact、IDE 桥、工具执行等
-│   ├── context_system/     # workspace / git / CLAUDE.md 上下文
+│   ├── context_system/     # workspace / git / CLAWCODEX.md 上下文
 │   ├── permissions/        # 权限模式与 bash 解析
 │   ├── hooks/              # Hook 类型与执行辅助
 │   └── command_system/     # 斜杠命令与参数替换

src/agent/agent_definitions.py

@@ -30,7 +30,7 @@ class AgentDefinition:
     background: bool = False
     color: str | None = None
     memory: str | None = None
-    omit_claude_md: bool = False
+    omit_clawcodex_md: bool = False
     disallowed_tools: list[str] | None = None
     hooks: dict[str, Any] | None = None
     skills: list[str] | None = None
@@ -157,7 +157,7 @@ def _explore_system_prompt(**_kwargs: Any) -> str:
     disallowed_tools=["Agent", "ExitPlanMode", "Edit", "Write", "NotebookEdit"],
     source="built-in",
     base_dir="built-in",
-    omit_claude_md=True,
+    omit_clawcodex_md=True,
     # ch08 round-4 (critic M1) — Explore is the fast/cheap read-only agent;
     # TS exploreAgent.ts:77 runs it on Haiku. get_agent_model resolves this
     # against the session provider and inherits on providers that don't
@@ -228,7 +228,7 @@ def _plan_system_prompt(**_kwargs: Any) -> str:
     source="built-in",
     base_dir="built-in",
     model="inherit",
-    omit_claude_md=True,
+    omit_clawcodex_md=True,
     get_system_prompt=_plan_system_prompt,
 )
 

src/agent/parse_agent_markdown.py

@@ -16,8 +16,8 @@
     background              → background
     color                   → color
     memory                  → memory
-    omit-claude-md          → omit_claude_md
-    omitClaudeMd            → omit_claude_md (camelCase alias)
+    omit-clawcodex-md       → omit_clawcodex_md
+    omitClawcodexMd         → omit_clawcodex_md (camelCase alias)
     hooks                   → hooks
     skills                  → skills
     isolation               → isolation
@@ -182,7 +182,9 @@ def parse_agent_from_markdown(
     background = _parse_bool(_first(frontmatter, "background"))
     color = _parse_color(_first(frontmatter, "color"))
     memory = _parse_memory(_first(frontmatter, "memory"), file_path=file_path)
-    omit_claude_md = _parse_bool(_first(frontmatter, "omit-claude-md", "omitClaudeMd"))
+    omit_clawcodex_md = _parse_bool(
+        _first(frontmatter, "omit-clawcodex-md", "omitClawcodexMd")
+    )
     hooks = parse_hooks(_first(frontmatter, "hooks"), owner_name=f"agent {agent_type}")
     skills = parse_string_list(_first(frontmatter, "skills"))
     isolation = _parse_isolation(
@@ -214,7 +216,7 @@ def _get_system_prompt(**_kwargs: Any) -> str:
         background=background,
         color=color,
         memory=memory,
-        omit_claude_md=omit_claude_md,
+        omit_clawcodex_md=omit_clawcodex_md,
         disallowed_tools=disallowed_tools,
         hooks=hooks,
         skills=skills or None,

src/bootstrap/state.py

@@ -203,10 +203,10 @@ class _BootstrapState:
     invoked_skills: dict[str, InvokedSkillInfo] = field(default_factory=dict)
 
     # --- Cache optimization (TS: lines 122-123, 202-205, 207, 256) ---------
-    cached_claude_md_content: str | None = None
+    cached_clawcodex_md_content: str | None = None
     system_prompt_section_cache: dict[str, str | None] = field(default_factory=dict)
     pending_post_compaction: bool = False
-    additional_directories_for_claude_md: list[str] = field(default_factory=list)
+    additional_directories_for_clawcodex_md: list[str] = field(default_factory=list)
 
     # --- Model (TS: lines 68-70) -------------------------------------------
     main_loop_model_override: str | None = None
@@ -772,12 +772,12 @@ def set_cost_state_for_restore(
 # ===========================================================================
 
 
-def get_cached_claude_md_content() -> str | None:
-    return _STATE.cached_claude_md_content
+def get_cached_clawcodex_md_content() -> str | None:
+    return _STATE.cached_clawcodex_md_content
 
 
-def set_cached_claude_md_content(content: str | None) -> None:
-    _STATE.cached_claude_md_content = content
+def set_cached_clawcodex_md_content(content: str | None) -> None:
+    _STATE.cached_clawcodex_md_content = content
 
 
 def get_system_prompt_section_cache() -> dict[str, str | None]:
@@ -806,12 +806,12 @@ def consume_post_compaction() -> bool:
     return was
 
 
-def get_additional_directories_for_claude_md() -> list[str]:
-    return _STATE.additional_directories_for_claude_md
+def get_additional_directories_for_clawcodex_md() -> list[str]:
+    return _STATE.additional_directories_for_clawcodex_md
 
 
-def set_additional_directories_for_claude_md(directories: list[str]) -> None:
-    _STATE.additional_directories_for_claude_md = list(directories)
+def set_additional_directories_for_clawcodex_md(directories: list[str]) -> None:
+    _STATE.additional_directories_for_clawcodex_md = list(directories)
 
 
 # ===========================================================================
@@ -1266,15 +1266,15 @@ def reset_state_for_tests() -> None:
     "is_repl_bridge_active",
     "get_repl_bridge_handle",
     # Cache optimization
-    "get_cached_claude_md_content",
-    "set_cached_claude_md_content",
+    "get_cached_clawcodex_md_content",
+    "set_cached_clawcodex_md_content",
     "get_system_prompt_section_cache",
     "set_system_prompt_section_cache_entry",
     "clear_system_prompt_section_state",
     "mark_post_compaction",
     "consume_post_compaction",
-    "get_additional_directories_for_claude_md",
-    "set_additional_directories_for_claude_md",
+    "get_additional_directories_for_clawcodex_md",
+    "set_additional_directories_for_clawcodex_md",
     # Model
     "get_main_loop_model_override",
     "set_main_loop_model_override",