klaude-code 1.2.17__py3-none-any.whl → 1.2.19__py3-none-any.whl

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/protocol/commands.py

@@ -3,6 +3,7 @@ from enum import Enum
 
 class CommandName(str, Enum):
     INIT = "init"
+    DEBUG = "debug"
     DIFF = "diff"
     HELP = "help"
     MODEL = "model"

klaude_code/protocol/sub_agent/web.py

@@ -20,14 +20,15 @@ Capabilities:
 
 How to use:
 - Write a clear prompt describing what information you need - the agent will search and fetch as needed
+- Account for "Today's date" in <env>. For example, if <env> says "Today's date: 2025-07-01", and the user wants the latest docs, do not use 2024 in the search query. Use 2025.
 - Optionally provide a `url` if you already know the target page
 - Use `output_format` (JSON Schema) to get structured data back from the agent
-- Account for "Today's date" in <env>. For example, if <env> says "Today's date: 2025-07-01", and the user wants the latest docs, do not use 2024 in the search query. Use 2025.
 
 What you receive:
 - The agent returns a text response summarizing its findings
 - With `output_format`, you receive structured JSON matching your schema
-- The response is the agent's analysis, not raw web content\
+- The response is the agent's analysis, not raw web content
+- Web content is saved to local files (paths included in Sources) - read them directly if you need full content\
 """
 
 WEB_AGENT_PARAMETERS = {

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())
@@ -36,7 +39,7 @@ class Session(BaseModel):
 
     @property
     def messages_count(self) -> int:
-        """Count of user and assistant messages in conversation history.
+        """Count of user, assistant messages, and tool calls in conversation history.
 
         This is a cached property that is invalidated when append_history is called.
         """
@@ -44,7 +47,7 @@ class Session(BaseModel):
             self._messages_count_cache = sum(
                 1
                 for it in self.conversation_history
-                if isinstance(it, (model.UserMessageItem, model.AssistantMessageItem))
+                if isinstance(it, (model.UserMessageItem, model.AssistantMessageItem, model.ToolCallItem))
             )
         return self._messages_count_cache
 
@@ -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

@@ -6,42 +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/geist-sans/latin-400.css"
+      href="https://cdn.jsdelivr.net/npm/@fontsource/geist/400.css"
       rel="stylesheet"
     />
     <link
-      href="https://cdn.jsdelivr.net/npm/@fontsource/geist-sans/latin-500.css"
+      href="https://cdn.jsdelivr.net/npm/@fontsource/geist/800.css"
       rel="stylesheet"
     />
     <link
-      href="https://cdn.jsdelivr.net/npm/@fontsource/geist-sans/latin-700.css"
+      href="https://cdn.jsdelivr.net/npm/@fontsource/jetbrains-mono/400.css"
+      rel="stylesheet"
+    />
+    <link
+      href="https://cdn.jsdelivr.net/npm/@fontsource/jetbrains-mono/800.css"
       rel="stylesheet"
     />
     <style>
       :root {
-        --bg-body: #ededed;
-        --bg-container: #f0f0f0;
-        --bg-card: #f0f0f0;
-        --border: #c8c8c8;
-        --text: #111111;
-        --text-dim: #64748b;
-        --accent: #0851b2;
-        --accent-dim: rgba(8, 145, 178, 0.08);
-        --success: #15803d;
-        --error: #dc2626;
-        --bg-error: #ffebee;
-        --bg-code: #f3f3f3;
-        --fg-inline-code: #4f4fc7;
-        --font-mono: ui-monospace, SFMono-Regular, Menlo, Monaco, Consolas, Liberation Mono, Courier New, monospace;
-        --font-markdown-mono: ui-monospace, SFMono-Regular, Menlo, Monaco, Consolas, Liberation Mono, Courier New, monospace;
-        --font-markdown: "Geist Sans", system-ui, sans-serif;
+        --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: 13px;
-        --font-size-sm: 14px;
-        --font-size-base: 15px;
+        --font-size-xs: 12px;
+        --font-size-sm: 13px;
+        --font-size-base: 14px;
         --font-size-lg: 16px;
         --spacing-xs: 4px;
         --spacing-sm: 8px;
@@ -62,7 +70,6 @@
         background-color: var(--bg-body);
         color: var(--text);
         font-family: var(--font-mono);
-        font-feature-settings: "ss18";
         line-height: 1.6;
         font-size: var(--font-size-lg);
         -webkit-font-smoothing: antialiased;
@@ -340,7 +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;
@@ -697,12 +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;
       }
 
@@ -813,13 +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 {
@@ -1267,10 +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>

klaude_code/trace/__init__.py

@@ -1,3 +1,21 @@
-from .log import DebugType, is_debug_enabled, log, log_debug, logger, prepare_debug_log_file, set_debug_logging
+from .log import (
+    DebugType,
+    get_current_log_file,
+    is_debug_enabled,
+    log,
+    log_debug,
+    logger,
+    prepare_debug_log_file,
+    set_debug_logging,
+)
 
-__all__ = ["DebugType", "is_debug_enabled", "log", "log_debug", "logger", "prepare_debug_log_file", "set_debug_logging"]
+__all__ = [
+    "DebugType",
+    "get_current_log_file",
+    "is_debug_enabled",
+    "log",
+    "log_debug",
+    "logger",
+    "prepare_debug_log_file",
+    "set_debug_logging",
+]