klaude-code 1.2.18__py3-none-any.whl → 1.2.19__py3-none-any.whl

klaude_code/core/tool/file/read_tool.py

@@ -209,8 +209,9 @@ class ReadTool(ToolABC):
             return model.ToolResultItem(
                 status="error",
                 output=(
-                    "<tool_use_error>PDF files are not supported by this tool. "
-                    "Please use a Python script with `pdfplumber` to extract text/tables:\n\n"
+                    "<tool_use_error>PDF files are not supported by this tool.\n"
+                    "If there's an available skill for PDF, use it.\n"
+                    "Or use a Python script with `pdfplumber` to extract text/tables:\n\n"
                     "```python\n"
                     "# /// script\n"
                     '# dependencies = ["pdfplumber"]\n'

klaude_code/core/tool/memory/skill_loader.py

@@ -132,20 +132,22 @@ class SkillLoader:
         for user_dir in self.USER_SKILLS_DIRS:
             expanded_dir = user_dir.expanduser()
             if expanded_dir.exists():
-                for skill_file in expanded_dir.rglob("SKILL.md"):
-                    skill = self.load_skill(skill_file, location="user")
-                    if skill:
-                        skills.append(skill)
-                        self.loaded_skills[skill.name] = skill
+                for pattern in ("SKILL.md", "skill.md"):
+                    for skill_file in expanded_dir.rglob(pattern):
+                        skill = self.load_skill(skill_file, location="user")
+                        if skill:
+                            skills.append(skill)
+                            self.loaded_skills[skill.name] = skill
 
         # Load project-level skills (override user skills if same name)
         project_dir = self.PROJECT_SKILLS_DIR.resolve()
         if project_dir.exists():
-            for skill_file in project_dir.rglob("SKILL.md"):
-                skill = self.load_skill(skill_file, location="project")
-                if skill:
-                    skills.append(skill)
-                    self.loaded_skills[skill.name] = skill
+            for pattern in ("SKILL.md", "skill.md"):
+                for skill_file in project_dir.rglob(pattern):
+                    skill = self.load_skill(skill_file, location="project")
+                    if skill:
+                        skills.append(skill)
+                        self.loaded_skills[skill.name] = skill
 
         # Log discovery summary
         if skills:

klaude_code/core/tool/tool_registry.py

@@ -66,7 +66,7 @@ def load_agent_tools(
 
     # Main agent tools
     if "gpt-5" in model_name:
-        tool_names = [tools.BASH, tools.APPLY_PATCH, tools.UPDATE_PLAN]
+        tool_names = [tools.BASH, tools.READ, tools.APPLY_PATCH, tools.UPDATE_PLAN]
     elif "gemini-3" in model_name:
         tool_names = [tools.BASH, tools.READ, tools.EDIT, tools.WRITE]
     else:

klaude_code/llm/anthropic/client.py

@@ -1,4 +1,5 @@
 import json
+import os
 from collections.abc import AsyncGenerator
 from typing import override
 
@@ -61,11 +62,20 @@ def build_payload(param: llm_param.LLMCallParameter) -> MessageCreateParamsStrea
 class AnthropicClient(LLMClientABC):
     def __init__(self, config: llm_param.LLMConfigParameter):
         super().__init__(config)
-        client = anthropic.AsyncAnthropic(
-            api_key=config.api_key,
-            base_url=config.base_url,
-            timeout=httpx.Timeout(300.0, connect=15.0, read=285.0),
-        )
+        # Remove ANTHROPIC_AUTH_TOKEN env var to prevent anthropic SDK from adding
+        # Authorization: Bearer header that may conflict with third-party APIs
+        # (e.g., deepseek, moonshot) that use Authorization header for authentication.
+        # The API key will be sent via X-Api-Key header instead.
+        saved_auth_token = os.environ.pop("ANTHROPIC_AUTH_TOKEN", None)
+        try:
+            client = anthropic.AsyncAnthropic(
+                api_key=config.api_key,
+                base_url=config.base_url,
+                timeout=httpx.Timeout(300.0, connect=15.0, read=285.0),
+            )
+        finally:
+            if saved_auth_token is not None:
+                os.environ["ANTHROPIC_AUTH_TOKEN"] = saved_auth_token
         self.client: anthropic.AsyncAnthropic = client
 
     @classmethod
@@ -120,35 +130,38 @@ class AnthropicClient(LLMClientABC):
                     case BetaRawContentBlockDeltaEvent() as event:
                         match event.delta:
                             case BetaThinkingDelta() as delta:
-                                metadata_tracker.record_token()
+                                if delta.thinking:
+                                    metadata_tracker.record_token()
                                 accumulated_thinking.append(delta.thinking)
                                 yield model.ReasoningTextDelta(
                                     content=delta.thinking,
                                     response_id=response_id,
                                 )
                             case BetaSignatureDelta() as delta:
-                                metadata_tracker.record_token()
                                 yield model.ReasoningEncryptedItem(
                                     encrypted_content=delta.signature,
                                     response_id=response_id,
                                     model=str(param.model),
                                 )
                             case BetaTextDelta() as delta:
-                                metadata_tracker.record_token()
+                                if delta.text:
+                                    metadata_tracker.record_token()
                                 accumulated_content.append(delta.text)
                                 yield model.AssistantMessageDelta(
                                     content=delta.text,
                                     response_id=response_id,
                                 )
                             case BetaInputJSONDelta() as delta:
-                                metadata_tracker.record_token()
                                 if current_tool_inputs is not None:
+                                    if delta.partial_json:
+                                        metadata_tracker.record_token()
                                     current_tool_inputs.append(delta.partial_json)
                             case _:
                                 pass
                     case BetaRawContentBlockStartEvent() as event:
                         match event.content_block:
                             case BetaToolUseBlock() as block:
+                                metadata_tracker.record_token()
                                 yield model.ToolCallStartItem(
                                     response_id=response_id,
                                     call_id=block.id,
@@ -161,6 +174,7 @@ class AnthropicClient(LLMClientABC):
                                 pass
                     case BetaRawContentBlockStopEvent() as event:
                         if len(accumulated_thinking) > 0:
+                            metadata_tracker.record_token()
                             full_thinking = "".join(accumulated_thinking)
                             yield model.ReasoningTextItem(
                                 content=full_thinking,
@@ -169,12 +183,14 @@ class AnthropicClient(LLMClientABC):
                             )
                             accumulated_thinking.clear()
                         if len(accumulated_content) > 0:
+                            metadata_tracker.record_token()
                             yield model.AssistantMessageItem(
                                 content="".join(accumulated_content),
                                 response_id=response_id,
                             )
                             accumulated_content.clear()
                         if current_tool_name and current_tool_call_id:
+                            metadata_tracker.record_token()
                             yield model.ToolCallItem(
                                 name=current_tool_name,
                                 call_id=current_tool_call_id,

klaude_code/llm/openai_compatible/client.py

@@ -23,7 +23,7 @@ def build_payload(param: llm_param.LLMCallParameter) -> tuple[CompletionCreatePa
 
     extra_body: dict[str, object] = {}
 
-    if param.thinking:
+    if param.thinking and param.thinking.type == "enabled":
         extra_body["thinking"] = {
             "type": param.thinking.type,
             "budget": param.thinking.budget_tokens,
@@ -182,7 +182,10 @@ class OpenAICompatibleClient(LLMClientABC):
             yield model.StreamErrorItem(error=f"{e.__class__.__name__} {e!s}")
 
         # Finalize
-        for item in state.flush_all():
+        flushed_items = state.flush_all()
+        if flushed_items:
+            metadata_tracker.record_token()
+        for item in flushed_items:
             yield item
 
         metadata_tracker.set_response_id(state.response_id)

klaude_code/llm/openrouter/client.py

@@ -35,7 +35,7 @@ def build_payload(
     extra_headers: dict[str, str] = {}
 
     if param.thinking:
-        if param.thinking.budget_tokens is not None:
+        if param.thinking.type != "disabled" and param.thinking.budget_tokens is not None:
             extra_body["reasoning"] = {
                 "max_tokens": param.thinking.budget_tokens,
                 "enable": True,
@@ -139,7 +139,8 @@ class OpenRouterClient(LLMClientABC):
                     for item in reasoning_details:
                         try:
                             reasoning_detail = ReasoningDetail.model_validate(item)
-                            metadata_tracker.record_token()
+                            if reasoning_detail.text or reasoning_detail.summary:
+                                metadata_tracker.record_token()
                             state.stage = "reasoning"
                             # Yield delta immediately for streaming
                             if reasoning_detail.text:
@@ -198,7 +199,10 @@ class OpenRouterClient(LLMClientABC):
             yield model.StreamErrorItem(error=f"{e.__class__.__name__} {e!s}")
 
         # Finalize
-        for item in state.flush_all():
+        flushed_items = state.flush_all()
+        if flushed_items:
+            metadata_tracker.record_token()
+        for item in flushed_items:
             yield item
 
         metadata_tracker.set_response_id(state.response_id)

klaude_code/llm/responses/client.py

@@ -77,6 +77,7 @@ async def parse_responses_stream(
                     yield model.StartItem(response_id=response_id)
                 case responses.ResponseReasoningSummaryTextDeltaEvent() as event:
                     if event.delta:
+                        metadata_tracker.record_token()
                         yield model.ReasoningTextDelta(
                             content=event.delta,
                             response_id=response_id,
@@ -89,10 +90,12 @@ async def parse_responses_stream(
                             model=str(param.model),
                         )
                 case responses.ResponseTextDeltaEvent() as event:
-                    metadata_tracker.record_token()
+                    if event.delta:
+                        metadata_tracker.record_token()
                     yield model.AssistantMessageDelta(content=event.delta, response_id=response_id)
                 case responses.ResponseOutputItemAddedEvent() as event:
                     if isinstance(event.item, responses.ResponseFunctionToolCall):
+                        metadata_tracker.record_token()
                         yield model.ToolCallStartItem(
                             response_id=response_id,
                             call_id=event.item.call_id,
@@ -102,6 +105,7 @@ async def parse_responses_stream(
                     match event.item:
                         case responses.ResponseReasoningItem() as item:
                             if item.encrypted_content:
+                                metadata_tracker.record_token()
                                 yield model.ReasoningEncryptedItem(
                                     id=item.id,
                                     encrypted_content=item.encrypted_content,
@@ -109,6 +113,7 @@ async def parse_responses_stream(
                                     model=str(param.model),
                                 )
                         case responses.ResponseOutputMessage() as item:
+                            metadata_tracker.record_token()
                             yield model.AssistantMessageItem(
                                 content="\n".join(
                                     [

klaude_code/session/session.py

@@ -7,7 +7,7 @@ from typing import ClassVar
 
 from pydantic import BaseModel, Field, PrivateAttr
 
-from klaude_code.protocol import events, model, tools
+from klaude_code.protocol import events, llm_param, model, tools
 
 
 class Session(BaseModel):
@@ -22,6 +22,9 @@ class Session(BaseModel):
     # Model name used for this session
     # Used in list method SessionMetaBrief
     model_name: str | None = None
+
+    model_config_name: str | None = None
+    model_thinking: llm_param.Thinking | None = None
     # Timestamps (epoch seconds)
     created_at: float = Field(default_factory=lambda: time.time())
     updated_at: float = Field(default_factory=lambda: time.time())
@@ -103,14 +106,7 @@ class Session(BaseModel):
         return self._messages_dir() / f"{prefix}-{self.id}.jsonl"
 
     @classmethod
-    def create(cls, id: str | None = None) -> "Session":
-        """Create a new session without checking for existing files."""
-        return Session(id=id or uuid.uuid4().hex, work_dir=Path.cwd())
-
-    @classmethod
-    def load(cls, id: str) -> "Session":
-        """Load an existing session or create a new one if not found."""
-        # Load session metadata
+    def _find_session_file(cls, id: str) -> Path | None:
         sessions_dir = cls._sessions_dir()
         session_candidates = sorted(
             sessions_dir.glob(f"*-{id}.json"),
@@ -118,13 +114,24 @@ class Session(BaseModel):
             reverse=True,
         )
         if not session_candidates:
-            # No existing session; create a new one
+            return None
+        return session_candidates[0]
+
+    @classmethod
+    def create(cls, id: str | None = None) -> "Session":
+        """Create a new session without checking for existing files."""
+        return Session(id=id or uuid.uuid4().hex, work_dir=Path.cwd())
+
+    @classmethod
+    def load_meta(cls, id: str) -> "Session":
+        """Load session metadata only (without loading messages history)."""
+
+        session_path = cls._find_session_file(id)
+        if session_path is None:
             return Session(id=id, work_dir=Path.cwd())
-        session_path = session_candidates[0]
 
         raw = json.loads(session_path.read_text())
 
-        # Basic fields (conversation history is loaded separately)
         work_dir_str = raw.get("work_dir", str(Path.cwd()))
 
         sub_agent_state_raw = raw.get("sub_agent_state")
@@ -135,8 +142,12 @@ class Session(BaseModel):
         created_at = float(raw.get("created_at", time.time()))
         updated_at = float(raw.get("updated_at", created_at))
         model_name = raw.get("model_name")
+        model_config_name = raw.get("model_config_name")
 
-        sess = Session(
+        model_thinking_raw = raw.get("model_thinking")
+        model_thinking = llm_param.Thinking(**model_thinking_raw) if isinstance(model_thinking_raw, dict) else None  # pyright: ignore[reportUnknownArgumentType]
+
+        return Session(
             id=id,
             work_dir=Path(work_dir_str),
             sub_agent_state=sub_agent_state,
@@ -146,8 +157,15 @@ class Session(BaseModel):
             created_at=created_at,
             updated_at=updated_at,
             model_name=model_name,
+            model_config_name=model_config_name,
+            model_thinking=model_thinking,
         )
 
+    @classmethod
+    def load(cls, id: str) -> "Session":
+        """Load an existing session or create a new one if not found."""
+        sess = cls.load_meta(id)
+
         # Load conversation history from messages JSONL
         messages_dir = cls._messages_dir()
         # Expect a single messages file per session (prefixed filenames only)
@@ -204,6 +222,8 @@ class Session(BaseModel):
             "updated_at": self.updated_at,
             "messages_count": self.messages_count,
             "model_name": self.model_name,
+            "model_config_name": self.model_config_name,
+            "model_thinking": self.model_thinking.model_dump() if self.model_thinking else None,
         }
         self._session_file().write_text(json.dumps(payload, ensure_ascii=False, indent=2))
 

klaude_code/session/templates/export_session.html

@@ -6,54 +6,50 @@
     <title>Klaude Code - $first_user_message</title>
     <link
       rel="icon"
-      href="data:image/svg+xml,<svg xmlns=%22http://www.w3.org/2000/svg%22 viewBox=%220 0 24 24%22 fill=%22none%22 stroke=%22%230851b2%22 stroke-width=%222%22 stroke-linecap=%22round%22 stroke-linejoin=%22round%22><polyline points=%2216 18 22 12 16 6%22></polyline><polyline points=%228 6 2 12 8 18%22></polyline></svg>"
+      href="data:image/svg+xml,<svg xmlns=%22http://www.w3.org/2000/svg%22 viewBox=%220 0 24 24%22 fill=%22none%22 stroke=%22%230b5bd3%22 stroke-width=%222%22 stroke-linecap=%22round%22 stroke-linejoin=%22round%22><polyline points=%2216 18 22 12 16 6%22></polyline><polyline points=%228 6 2 12 8 18%22></polyline></svg>"
     />
     <link
-      href="https://cdn.jsdelivr.net/npm/@fontsource/source-sans-3/400.css"
+      href="https://cdn.jsdelivr.net/npm/@fontsource/geist/400.css"
       rel="stylesheet"
     />
     <link
-      href="https://cdn.jsdelivr.net/npm/@fontsource/source-sans-3/400-italic.css"
+      href="https://cdn.jsdelivr.net/npm/@fontsource/geist/800.css"
       rel="stylesheet"
     />
     <link
-      href="https://cdn.jsdelivr.net/npm/@fontsource/source-sans-3/700.css"
+      href="https://cdn.jsdelivr.net/npm/@fontsource/jetbrains-mono/400.css"
       rel="stylesheet"
     />
     <link
-      href="https://cdn.jsdelivr.net/npm/@fontsource/source-sans-3/700-italic.css"
-      rel="stylesheet"
-    />
-    <link
-      href="https://cdn.jsdelivr.net/npm/@fontsource/fira-code/400.css"
-      rel="stylesheet"
-    />
-    <link
-      href="https://cdn.jsdelivr.net/npm/@fontsource/fira-code/700.css"
+      href="https://cdn.jsdelivr.net/npm/@fontsource/jetbrains-mono/800.css"
       rel="stylesheet"
     />
     <style>
       :root {
-        --bg-body: #eae9e5;
-        --bg-container: #edece9;
-        --bg-card: #efeeeb;
-        --bg-error: #ffebee;
-        --bg-code: #f2f1ed;
-        --border: #c8c8c8;
-        --text: #111111;
-        --text-dim: #64748b;
-        --accent: #0851b2;
-        --accent-dim: rgba(8, 145, 178, 0.08);
-        --success: #15803d;
-        --error: #dc2626;
-        --fg-inline-code: #4f4fc7;
-        --font-mono: "Fira Code", ui-monospace, SFMono-Regular, Menlo, Monaco, Consolas, Liberation Mono, Courier New, monospace;
-        --font-markdown-mono: "Fira Code", ui-monospace, SFMono-Regular, Menlo, Monaco, Consolas, Liberation Mono, Courier New, monospace;
-        --font-markdown: "Source Sans 3", system-ui, sans-serif;
-        --font-weight-bold: 700;
-        --font-size-xs: 13px;
-        --font-size-sm: 14px;
-        --font-size-base: 15px;
+        --bg-body: #f0efeb;
+        --bg-container: #f3f2ee;
+        --bg-card: #f3f2ee;
+        --bg-overlay: rgba(248, 246, 240, 0.98);
+        --bg-error: #fdecec;
+        --bg-success: #eaf6ed;
+        --bg-code: #f7f7f4;
+        --border: #ded8cf;
+        --text: #151515;
+        --text-dim: #5f6b7a;
+        --accent: #0b5bd3;
+        --accent-dim: rgba(11, 91, 211, 0.10);
+        --accent-underline: rgba(11, 91, 211, 0.25);
+        --accent-hover: rgba(11, 91, 211, 0.06);
+        --success: #1a7f37;
+        --error: #c62828;
+        --fg-inline-code: #1f4fbf;
+        --font-mono: "JetBrains Mono", ui-monospace, SFMono-Regular, Menlo, Monaco, Consolas, Liberation Mono, Courier New, monospace;
+        --font-markdown-mono: "JetBrains Mono", ui-monospace, SFMono-Regular, Menlo, Monaco, Consolas, Liberation Mono, Courier New, monospace;
+        --font-markdown: "JetBrains Mono", ui-monospace, SFMono-Regular, Menlo, Monaco, Consolas, Liberation Mono, Courier New, monospace;
+        --font-weight-bold: 800;
+        --font-size-xs: 12px;
+        --font-size-sm: 13px;
+        --font-size-base: 14px;
         --font-size-lg: 16px;
         --spacing-xs: 4px;
         --spacing-sm: 8px;
@@ -351,7 +347,7 @@
         left: 0;
         width: 100vw;
         height: 100vh;
-        background: rgba(255, 255, 255, 0.98);
+        background: var(--bg-overlay);
         z-index: 1000;
         display: flex;
         flex-direction: column;
@@ -708,12 +704,12 @@
       .markdown-body a {
         color: var(--accent);
         text-decoration: none;
-        border-bottom: 1px solid rgba(8, 81, 178, 0.2);
+        border-bottom: 1px solid var(--accent-underline);
         transition: border-color 0.2s, background-color 0.2s;
       }
       .markdown-body a:hover {
         border-bottom-color: var(--accent);
-        background-color: rgba(8, 81, 178, 0.05);
+        background-color: var(--accent-hover);
         border-radius: 2px;
       }
 
@@ -824,13 +820,13 @@
         white-space: pre;
       }
       .diff-plus {
-        color: #166534;
-        background: #e7f5e9;
+        color: var(--success);
+        background: var(--bg-success);
         display: block;
       }
       .diff-minus {
-        color: #991b1b;
-        background: #ffebee;
+        color: var(--error);
+        background: var(--bg-error);
         display: block;
       }
       .diff-ctx {
@@ -1278,10 +1274,16 @@
     <script src="https://cdn.jsdelivr.net/npm/marked/marked.min.js"></script>
     <script type="module">
       import mermaid from "https://cdn.jsdelivr.net/npm/mermaid@10/dist/mermaid.esm.min.mjs";
+      const markdownMonoFont = getComputedStyle(document.documentElement)
+        .getPropertyValue("--font-markdown-mono")
+        .trim();
+
       mermaid.initialize({
         startOnLoad: true,
         theme: "neutral",
-        fontFamily: 'ui-monospace, SFMono-Regular, Menlo, Monaco, Consolas, Liberation Mono, Courier New, monospace',
+        fontFamily:
+          markdownMonoFont ||
+          'ui-monospace, SFMono-Regular, Menlo, Monaco, Consolas, Liberation Mono, Courier New, monospace',
       });
     </script>
     <script>