klaude-code 1.2.16__tar.gz → 1.2.18__tar.gz

{klaude_code-1.2.16 → klaude_code-1.2.18}/PKG-INFO

@@ -1,8 +1,9 @@
 Metadata-Version: 2.3
 Name: klaude-code
-Version: 1.2.16
+Version: 1.2.18
 Summary: Add your description here
 Requires-Dist: anthropic>=0.66.0
+Requires-Dist: ddgs>=9.9.3
 Requires-Dist: openai>=1.102.0
 Requires-Dist: pillow>=12.0.0
 Requires-Dist: prompt-toolkit>=3.0.52
@@ -50,12 +51,17 @@ klaude [--model <name>] [--select-model]
 
 **Options:**
 - `--version`/`-V`: Show version and exit.
-- `--model`/`-m`: Select a model by logical name from config.
-- `--select-model`/`-s`: Interactively choose a model at startup.
+- `--model`/`-m`: Preferred model name (exact match picks immediately; otherwise opens the interactive selector filtered by this value).
+- `--select-model`/`-s`: Open the interactive model selector at startup (shows all models unless `--model` is also provided).
 - `--continue`/`-c`: Resume the most recent session.
 - `--resume`/`-r`: Select a session to resume for this project.
 - `--vanilla`: Minimal mode with only basic tools (Bash, Read, Edit) and no system prompts.
 
+**Model selection behavior:**
+- Default: uses `main_model` from config.
+- `--select-model`: always prompts you to pick.
+- `--model <value>`: tries to resolve `<value>` to a single model; if it can't, it prompts with a filtered list (and falls back to showing all models if there are no matches).
+
 **Debug Options:**
 - `--debug`/`-d`: Enable debug mode with verbose logging and LLM trace.
 - `--debug-filter`: Filter debug output by type (comma-separated).
@@ -63,7 +69,7 @@ klaude [--model <name>] [--select-model]
 
 ### Configuration
 
-An example config will be created in `~/.klaude/config.yaml` when first run.
+An example config will be created in `~/.klaude/klaude-config.yaml` when first run.
 
 Open the configuration file in editor:
 
@@ -71,47 +77,137 @@ Open the configuration file in editor:
 klaude config
 ```
 
-An minimal example config yaml using OpenRouter's API Key:
+An example config yaml:
 
 ```yaml
 provider_list:
-- provider_name: openrouter-work
+- provider_name: openrouter
   protocol: openrouter # support <responses|openrouter|anthropic|openai>
   api_key: <your-openrouter-api-key>
+
+- provider_name: openai-responses
+  protocol: responses
+  api_key: <your-openai-api-key>
+
+- provider_name: anthropic
+  protocol: anthropic
+  api_key: <your-anthropic-api-key>
+
+- provider_name: moonshot
+  protocol: anthropic
+  base_url: https://api.moonshot.cn/anthropic
+  api_key: <your-api-key>
+
+- provider_name: deepseek
+  protocol: anthropic
+  base_url: https://api.deepseek.com/anthropic
+  api_key: <your-api-key>
+
 model_list:
-- model_name: gpt-5.1-codex
-  provider: openrouter
+
+- model_name: deepseek
+  provider: deepseek
+  model_params:
+    model: deepseek-reasoner
+    context_limit: 128000
+    thinking:
+      type: enabled
+      budget_tokens: 8192
+    cost:
+      currency: CNY
+      input: 2
+      output: 3
+      cache_read: 0.2
+
+- model_name: codex-max
+  provider: openai-responses
   model_params:
-    model: openai/gpt-5.1-codex
-    context_limit: 368000
+    model: gpt-5.1-codex-max
     thinking:
       reasoning_effort: medium
-- model_name: gpt-5.1-high
+    context_limit: 400000
+    max_tokens: 128000
+    cost:
+      input: 1.25
+      output: 10
+      cache_read: 0.13
+
+- model_name: gpt-5.1
   provider: openrouter
   model_params:
     model: openai/gpt-5.1
-    context_limit: 368000
+    context_limit: 400000
+    max_tokens: 128000
+    verbosity: high
     thinking:
       reasoning_effort: high
-- model_name: sonnet
+    cost:
+      input: 1.25
+      output: 10
+      cache_read: 0.13
+
+- model_name: kimi@moonshot
+  provider: moonshot
+  model_params:
+    model: kimi-k2-thinking
+    context_limit: 262144
+    thinking:
+      type: enabled
+      budget_tokens: 8192
+    cost:
+      currency: CNY
+      input: 4
+      output: 16
+      cache_read: 1
+
+- model_name: opus
   provider: openrouter
   model_params:
-    model: anthropic/claude-4.5-sonnet
-    context_limit: 168000
+    model: anthropic/claude-4.5-opus
+    context_limit: 200000
     provider_routing:
-      sort: throughput
-- model_name: haiku
+      only: [ google-vertex ]
+    verbosity: high
+    thinking:
+      type: enabled
+      budget_tokens: 31999
+    cost:
+      input: 5
+      output: 25
+      cache_read: 0.5
+      cache_write: 6.25
+
+- model_name: gemini
   provider: openrouter
   model_params:
-    model: anthropic/claude-haiku-4.5
-    context_limit: 168000
-    provider_routing:
-      sort: throughput
-main_model: gpt-5.1-codex
+    model: google/gemini-3-pro-preview
+    context_limit: 1048576
+    thinking:
+      reasoning_effort: medium
+    cost:
+      input: 2
+      output: 12
+      cache_read: 0.2
+
+- model_name: haiku
+  provider: anthropic
+  model_params:
+    model: claude-haiku-4-5-20251001
+    context_limit: 200000
+    cost:
+      input: 1
+      output: 5
+      cache_read: 0.1
+      cache_write: 1.25
+
+main_model: opus
+
 sub_agent_models:
-  oracle: gpt-5.1-high
+  oracle: gpt-5.1
   explore: haiku
-  task: sonnet
+  task: opus
+  webagent: haiku
+
 ```
 
 List configured providers and models:
@@ -172,5 +268,10 @@ klaude exec "what is 2+2?"
 echo "hello world" | klaude exec
 
 # With model selection
+
+# Exact model name (non-interactive)
 echo "generate quicksort in python" | klaude exec --model gpt-5.1
-```
+
+# Partial/ambiguous name opens the interactive selector (filtered)
+echo "generate quicksort in python" | klaude exec --model gpt
+```

{klaude_code-1.2.16 → klaude_code-1.2.18}/README.md

@@ -33,12 +33,17 @@ klaude [--model <name>] [--select-model]
 
 **Options:**
 - `--version`/`-V`: Show version and exit.
-- `--model`/`-m`: Select a model by logical name from config.
-- `--select-model`/`-s`: Interactively choose a model at startup.
+- `--model`/`-m`: Preferred model name (exact match picks immediately; otherwise opens the interactive selector filtered by this value).
+- `--select-model`/`-s`: Open the interactive model selector at startup (shows all models unless `--model` is also provided).
 - `--continue`/`-c`: Resume the most recent session.
 - `--resume`/`-r`: Select a session to resume for this project.
 - `--vanilla`: Minimal mode with only basic tools (Bash, Read, Edit) and no system prompts.
 
+**Model selection behavior:**
+- Default: uses `main_model` from config.
+- `--select-model`: always prompts you to pick.
+- `--model <value>`: tries to resolve `<value>` to a single model; if it can't, it prompts with a filtered list (and falls back to showing all models if there are no matches).
+
 **Debug Options:**
 - `--debug`/`-d`: Enable debug mode with verbose logging and LLM trace.
 - `--debug-filter`: Filter debug output by type (comma-separated).
@@ -46,7 +51,7 @@ klaude [--model <name>] [--select-model]
 
 ### Configuration
 
-An example config will be created in `~/.klaude/config.yaml` when first run.
+An example config will be created in `~/.klaude/klaude-config.yaml` when first run.
 
 Open the configuration file in editor:
 
@@ -54,47 +59,137 @@ Open the configuration file in editor:
 klaude config
 ```
 
-An minimal example config yaml using OpenRouter's API Key:
+An example config yaml:
 
 ```yaml
 provider_list:
-- provider_name: openrouter-work
+- provider_name: openrouter
   protocol: openrouter # support <responses|openrouter|anthropic|openai>
   api_key: <your-openrouter-api-key>
+
+- provider_name: openai-responses
+  protocol: responses
+  api_key: <your-openai-api-key>
+
+- provider_name: anthropic
+  protocol: anthropic
+  api_key: <your-anthropic-api-key>
+
+- provider_name: moonshot
+  protocol: anthropic
+  base_url: https://api.moonshot.cn/anthropic
+  api_key: <your-api-key>
+
+- provider_name: deepseek
+  protocol: anthropic
+  base_url: https://api.deepseek.com/anthropic
+  api_key: <your-api-key>
+
 model_list:
-- model_name: gpt-5.1-codex
-  provider: openrouter
+
+- model_name: deepseek
+  provider: deepseek
+  model_params:
+    model: deepseek-reasoner
+    context_limit: 128000
+    thinking:
+      type: enabled
+      budget_tokens: 8192
+    cost:
+      currency: CNY
+      input: 2
+      output: 3
+      cache_read: 0.2
+
+- model_name: codex-max
+  provider: openai-responses
   model_params:
-    model: openai/gpt-5.1-codex
-    context_limit: 368000
+    model: gpt-5.1-codex-max
     thinking:
       reasoning_effort: medium
-- model_name: gpt-5.1-high
+    context_limit: 400000
+    max_tokens: 128000
+    cost:
+      input: 1.25
+      output: 10
+      cache_read: 0.13
+
+- model_name: gpt-5.1
   provider: openrouter
   model_params:
     model: openai/gpt-5.1
-    context_limit: 368000
+    context_limit: 400000
+    max_tokens: 128000
+    verbosity: high
     thinking:
       reasoning_effort: high
-- model_name: sonnet
+    cost:
+      input: 1.25
+      output: 10
+      cache_read: 0.13
+
+- model_name: kimi@moonshot
+  provider: moonshot
+  model_params:
+    model: kimi-k2-thinking
+    context_limit: 262144
+    thinking:
+      type: enabled
+      budget_tokens: 8192
+    cost:
+      currency: CNY
+      input: 4
+      output: 16
+      cache_read: 1
+
+- model_name: opus
   provider: openrouter
   model_params:
-    model: anthropic/claude-4.5-sonnet
-    context_limit: 168000
+    model: anthropic/claude-4.5-opus
+    context_limit: 200000
     provider_routing:
-      sort: throughput
-- model_name: haiku
+      only: [ google-vertex ]
+    verbosity: high
+    thinking:
+      type: enabled
+      budget_tokens: 31999
+    cost:
+      input: 5
+      output: 25
+      cache_read: 0.5
+      cache_write: 6.25
+
+- model_name: gemini
   provider: openrouter
   model_params:
-    model: anthropic/claude-haiku-4.5
-    context_limit: 168000
-    provider_routing:
-      sort: throughput
-main_model: gpt-5.1-codex
+    model: google/gemini-3-pro-preview
+    context_limit: 1048576
+    thinking:
+      reasoning_effort: medium
+    cost:
+      input: 2
+      output: 12
+      cache_read: 0.2
+
+- model_name: haiku
+  provider: anthropic
+  model_params:
+    model: claude-haiku-4-5-20251001
+    context_limit: 200000
+    cost:
+      input: 1
+      output: 5
+      cache_read: 0.1
+      cache_write: 1.25
+
+main_model: opus
+
 sub_agent_models:
-  oracle: gpt-5.1-high
+  oracle: gpt-5.1
   explore: haiku
-  task: sonnet
+  task: opus
+  webagent: haiku
+
 ```
 
 List configured providers and models:
@@ -155,5 +250,10 @@ klaude exec "what is 2+2?"
 echo "hello world" | klaude exec
 
 # With model selection
+
+# Exact model name (non-interactive)
 echo "generate quicksort in python" | klaude exec --model gpt-5.1
-```
+
+# Partial/ambiguous name opens the interactive selector (filtered)
+echo "generate quicksort in python" | klaude exec --model gpt
+```

{klaude_code-1.2.16 → klaude_code-1.2.18}/pyproject.toml

@@ -4,12 +4,13 @@ build-backend = "uv_build"
 
 [project]
 name = "klaude-code"
-version = "1.2.16"
+version = "1.2.18"
 description = "Add your description here"
 readme = "README.md"
 requires-python = ">=3.13"
 dependencies = [
     "anthropic>=0.66.0",
+    "ddgs>=9.9.3",
     "openai>=1.102.0",
     "pillow>=12.0.0",
     "prompt-toolkit>=3.0.52",
@@ -76,10 +77,14 @@ name = "Layered architecture"
 type = "layers"
 layers = [
     "klaude_code.cli",
+    "klaude_code.ui",
     "klaude_code.core",
+    "klaude_code.command",
     "klaude_code.session",
     "klaude_code.config",
     "klaude_code.llm",
     "klaude_code.protocol",
+    "klaude_code.trace",
+    "klaude_code.auth",
     "klaude_code.const",
 ]

{klaude_code-1.2.16 → klaude_code-1.2.18}/src/klaude_code/cli/config_cmd.py

@@ -12,7 +12,7 @@ from klaude_code.trace import log
 
 def list_models() -> None:
     """List all models and providers configuration"""
-    from klaude_code.config.list_model import display_models_and_providers
+    from klaude_code.cli.list_model import display_models_and_providers
     from klaude_code.ui.terminal.color import is_light_terminal_background
 
     config = load_config()

{klaude_code-1.2.16 → klaude_code-1.2.18}/src/klaude_code/cli/debug.py

@@ -48,7 +48,7 @@ def open_log_file_in_editor(path: Path) -> None:
     editor = os.environ.get("EDITOR")
 
     if not editor:
-        for cmd in ["open", "xdg-open", "code", "nvim", "vim", "nano"]:
+        for cmd in ["open", "xdg-open", "code", "TextEdit", "notepad"]:
             try:
                 subprocess.run(["which", cmd], check=True, capture_output=True)
                 editor = cmd

{klaude_code-1.2.16 → klaude_code-1.2.18}/src/klaude_code/cli/main.py

@@ -11,7 +11,6 @@ from klaude_code.cli.auth_cmd import register_auth_commands
 from klaude_code.cli.config_cmd import register_config_commands
 from klaude_code.cli.debug import DEBUG_FILTER_HELP, open_log_file_in_editor, resolve_debug_settings
 from klaude_code.cli.session_cmd import register_session_commands
-from klaude_code.config import load_config
 from klaude_code.session import Session, resume_select_session
 from klaude_code.trace import DebugType, prepare_debug_log_file
 
@@ -157,13 +156,8 @@ def exec_command(
     from klaude_code.config.select_model import select_model_from_config
 
     chosen_model = model
-    if select_model:
-        # Prefer the explicitly provided model as default; otherwise main model
-        config = load_config()
-        if config is None:
-            raise typer.Exit(1)
-        default_name = model or config.main_model
-        chosen_model = select_model_from_config(preferred=default_name)
+    if model or select_model:
+        chosen_model = select_model_from_config(preferred=model)
         if chosen_model is None:
             return
 
@@ -243,7 +237,7 @@ def main_callback(
         setup_terminal_title()
 
         chosen_model = model
-        if select_model:
+        if model or select_model:
             chosen_model = select_model_from_config(preferred=model)
             if chosen_model is None:
                 return

{klaude_code-1.2.16 → klaude_code-1.2.18}/src/klaude_code/cli/runtime.py

@@ -10,7 +10,7 @@ from rich.text import Text
 from klaude_code import ui
 from klaude_code.command import has_interactive_command
 from klaude_code.config import Config, load_config
-from klaude_code.core.agent import Agent, DefaultModelProfileProvider, VanillaModelProfileProvider
+from klaude_code.core.agent import DefaultModelProfileProvider, VanillaModelProfileProvider
 from klaude_code.core.executor import Executor
 from klaude_code.core.manager import build_llm_clients
 from klaude_code.protocol import events, op
@@ -98,8 +98,10 @@ async def initialize_app_components(init_config: AppInitConfig) -> AppComponents
     executor_task = asyncio.create_task(executor.start())
 
     theme: str | None = config.theme
-    if theme is None:
+    if theme is None and not init_config.is_exec_mode:
         # Auto-detect theme from terminal background when config does not specify a theme.
+        # Skip detection in exec mode to avoid TTY race conditions with parent process's
+        # ESC monitor when running as a subprocess.
         detected = is_light_terminal_background()
         if detected is True:
             theme = "light"
@@ -145,8 +147,8 @@ async def initialize_session(
     await executor.submit_and_wait(op.InitAgentOperation(session_id=session_id))
     await event_queue.join()
 
-    active_session_ids = executor.context.agent_manager.active_session_ids()
-    return active_session_ids[0] if active_session_ids else session_id
+    active_session_id = executor.context.current_session_id()
+    return active_session_id or session_id
 
 
 async def cleanup_app_components(components: AppComponents) -> None:
@@ -214,16 +216,12 @@ async def run_interactive(init_config: AppInitConfig, session_id: str | None = N
 
     # Create status provider for bottom toolbar
     def _status_provider() -> REPLStatusSnapshot:
-        agent: Agent | None = None
-        # Get the first active agent (there should only be one in interactive mode)
-        active_agents = components.executor.context.active_agents
-        if active_agents:
-            agent = next(iter(active_agents.values()), None)
-
         # Check for updates (returns None if uv not available)
         update_message = get_update_message()
 
-        return build_repl_status_snapshot(agent=agent, update_message=update_message)
+        return build_repl_status_snapshot(
+            agent=components.executor.context.current_agent, update_message=update_message
+        )
 
     # Set up input provider for interactive mode
     input_provider: ui.InputProviderABC = ui.PromptToolkitInput(status_provider=_status_provider)
@@ -261,7 +259,14 @@ async def run_interactive(init_config: AppInitConfig, session_id: str | None = N
     restore_sigint = install_sigint_double_press_exit(_show_toast_once, _hide_progress)
 
     try:
-        active_session_id = await initialize_session(components.executor, components.event_queue, session_id=session_id)
+        await initialize_session(components.executor, components.event_queue, session_id=session_id)
+
+        def _get_active_session_id() -> str | None:
+            """Get the current active session ID dynamically.
+
+            This is necessary because /clear command creates a new session with a different ID.
+            """
+            return components.executor.context.current_session_id()
 
         # Input
         await input_provider.start()
@@ -272,6 +277,8 @@ async def run_interactive(init_config: AppInitConfig, session_id: str | None = N
             elif user_input.text.strip() == "":
                 continue
             # Submit user input operation - directly use the payload from iter_inputs
+            # Use dynamic session_id lookup to handle /clear creating new sessions
+            active_session_id = _get_active_session_id()
             submission_id = await components.executor.submit(
                 op.UserInputOperation(input=user_input, session_id=active_session_id)
             )
@@ -282,7 +289,7 @@ async def run_interactive(init_config: AppInitConfig, session_id: str | None = N
             else:
                 # Esc monitor for long-running, interruptible operations
                 async def _on_esc_interrupt() -> None:
-                    await components.executor.submit(op.InterruptOperation(target_session_id=active_session_id))
+                    await components.executor.submit(op.InterruptOperation(target_session_id=_get_active_session_id()))
 
                 stop_event, esc_task = start_esc_interrupt_monitor(_on_esc_interrupt)
                 # Wait for this specific task to complete before accepting next input

{klaude_code-1.2.16 → klaude_code-1.2.18}/src/klaude_code/command/__init__.py

@@ -28,8 +28,10 @@ def ensure_commands_loaded() -> None:
 
     # Import and register commands in display order
     from .clear_cmd import ClearCommand
+    from .debug_cmd import DebugCommand
     from .diff_cmd import DiffCommand
     from .export_cmd import ExportCommand
+    from .export_online_cmd import ExportOnlineCommand
     from .help_cmd import HelpCommand
     from .model_cmd import ModelCommand
     from .refresh_cmd import RefreshTerminalCommand
@@ -40,16 +42,18 @@ def ensure_commands_loaded() -> None:
 
     # Register in desired display order
     register(ExportCommand())
+    register(ExportOnlineCommand())
     register(RefreshTerminalCommand())
     register(ThinkingCommand())
     register(ModelCommand())
     load_prompt_commands()
-    register(ClearCommand())
     register(StatusCommand())
     register(DiffCommand())
     register(HelpCommand())
     register(ReleaseNotesCommand())
     register(TerminalSetupCommand())
+    register(DebugCommand())
+    register(ClearCommand())
 
     # Load prompt-based commands (appended after built-in commands)
 
@@ -58,8 +62,10 @@ def ensure_commands_loaded() -> None:
 def __getattr__(name: str) -> object:
     _commands_map = {
         "ClearCommand": "clear_cmd",
+        "DebugCommand": "debug_cmd",
         "DiffCommand": "diff_cmd",
         "ExportCommand": "export_cmd",
+        "ExportOnlineCommand": "export_online_cmd",
         "HelpCommand": "help_cmd",
         "ModelCommand": "model_cmd",
         "RefreshTerminalCommand": "refresh_cmd",

{klaude_code-1.2.16 → klaude_code-1.2.18}/src/klaude_code/command/clear_cmd.py

@@ -1,11 +1,6 @@
-from typing import TYPE_CHECKING
-
-from klaude_code.command.command_abc import CommandABC, CommandResult, InputAction
+from klaude_code.command.command_abc import Agent, CommandABC, CommandResult, InputAction
 from klaude_code.protocol import commands
 
-if TYPE_CHECKING:
-    from klaude_code.core.agent import Agent
-
 
 class ClearCommand(CommandABC):
     """Clear current session and start a new conversation"""
@@ -18,5 +13,5 @@ class ClearCommand(CommandABC):
     def summary(self) -> str:
         return "Clear conversation history and free up context"
 
-    async def run(self, raw: str, agent: "Agent") -> CommandResult:
+    async def run(self, raw: str, agent: Agent) -> CommandResult:
         return CommandResult(actions=[InputAction.clear()])