kolega-code 0.8.4__tar.gz → 0.9.0__tar.gz

{kolega_code-0.8.4 → kolega_code-0.9.0}/.env.example

@@ -12,6 +12,7 @@ MOONSHOT_API_KEY=
 DEEPSEEK_API_KEY=
 ZAI_API_KEY=
 KIMI_CODING_API_KEY=
+OLLAMA_API_KEY=
 
 KOLEGA_CODE_THINKING_EFFORT=
 

{kolega_code-0.8.4 → kolega_code-0.9.0}/CHANGELOG.md

@@ -6,6 +6,16 @@ This project uses GitHub Releases for detailed generated release notes. This fil
 
 ## Unreleased
 
+## 0.9.0 - 2026-06-24
+
+### Added
+
+- Added Ollama Cloud as a supported provider, including its model catalog and reasoning-field capture.
+
+### Fixed
+
+- Clarified the disconnected state shown on first run when no provider is configured.
+
 ## 0.8.4 - 2026-06-24
 
 ### Added

{kolega_code-0.8.4 → kolega_code-0.9.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: kolega-code
-Version: 0.8.4
+Version: 0.9.0
 Summary: Local-first AI coding agent for the terminal
 Project-URL: Homepage, https://kolega-ai.github.io/kolega-code/
 Project-URL: Documentation, https://kolega-ai.github.io/kolega-code/

{kolega_code-0.8.4 → kolega_code-0.9.0}/docs/src/content/docs/configuration/environment-variables.md

@@ -46,12 +46,14 @@ provider or model.
 | `DEEPSEEK_API_KEY` | DeepSeek |
 | `ZAI_API_KEY` | Z.AI (GLM Coding Plan) |
 | `KIMI_CODING_API_KEY` | Kimi Coding Plan |
+| `OLLAMA_API_KEY` | Ollama Cloud |
 
 The local `llama` provider needs no key. The `zai` key authenticates against Z.AI's
 Anthropic-compatible endpoint (it is the key Z.AI also documents as `ANTHROPIC_AUTH_TOKEN`).
 The `kimi_coding` key authenticates against the Kimi Coding Plan's separate
 Anthropic-compatible endpoint (`https://api.kimi.com/coding/`), which is distinct from the
 standard Moonshot API used by the `moonshot` provider.
+The `OLLAMA_API_KEY` key authenticates against Ollama Cloud's direct API (`https://ollama.com/v1` for OpenAI-compatible requests).
 
 ## Model selection
 
@@ -120,6 +122,7 @@ MOONSHOT_API_KEY=
 DEEPSEEK_API_KEY=
 FIREWORKS_API_KEY=
 ANTHROPIC_API_KEY=
+OLLAMA_API_KEY=
 
 # Optional: choose models per role
 KOLEGA_CODE_PROVIDER=moonshot
@@ -130,6 +133,10 @@ KOLEGA_CODE_THINKING_EFFORT=auto
 # KOLEGA_CODE_PROVIDER=fireworks
 # KOLEGA_CODE_MODEL=accounts/fireworks/models/glm-5p2
 
+# Ollama Cloud example:
+# KOLEGA_CODE_PROVIDER=ollama_cloud
+# KOLEGA_CODE_MODEL=glm-5.2
+
 # Optional: web search
 KOLEGA_CODE_WEB_SEARCH_BACKEND=duckduckgo
 

{kolega_code-0.8.4 → kolega_code-0.9.0}/docs/src/content/docs/configuration/providers-and-models.mdx

@@ -27,6 +27,7 @@ The agent runtime recognizes these providers:
 | DeepSeek | `deepseek` | `DEEPSEEK_API_KEY` |
 | Z.AI (GLM Coding Plan) | `zai` | `ZAI_API_KEY` |
 | Kimi Coding Plan | `kimi_coding` | `KIMI_CODING_API_KEY` |
+| Ollama Cloud | `ollama_cloud` | `OLLAMA_API_KEY` |
 | Llama (local) | `llama` | _none — local model_ |
 
 <Aside type="tip">
@@ -45,6 +46,15 @@ with thinking effort `auto` (enabled) requests route to **K2.7 Code**, and with
 (disabled) they route to **K2.6**. Set `KIMI_CODING_API_KEY` to your Kimi Coding Plan key.
 </Aside>
 
+<Aside type="tip">
+**Ollama Cloud.** The `ollama_cloud` provider uses Ollama Cloud's direct
+OpenAI-compatible API at `https://ollama.com/v1` with `OLLAMA_API_KEY`. Kolega
+Code uses the direct API model IDs returned by Ollama, such as **`glm-5.2`** and
+**`gpt-oss:120b`**, rather than local Ollama aliases such as `:cloud` or
+`-cloud`. Ollama may deprecate or retire older cloud models, so the supported
+model catalog should be refreshed periodically from Ollama's model library/API.
+</Aside>
+
 <Aside type="tip">
 **Fireworks.ai.** The `fireworks` provider targets Fireworks' OpenAI-compatible
 serverless endpoint (`https://api.fireworks.ai/inference/v1`) with `FIREWORKS_API_KEY`.
@@ -164,6 +174,8 @@ models in the TUI.
 | Z.AI `glm-5.2` | `high`, `max` | `max` |
 | Z.AI `glm-5.1` | `auto`, `none` | `auto` |
 | Fireworks serverless reasoning models | `none`, `low`, `medium`, `high`, `max` | `medium` |
+| Ollama Cloud thinking models | `none`, `low`, `medium`, `high`, `max` | `medium` |
+| Ollama Cloud `gpt-oss:*` | `low`, `medium`, `high` | `medium` |
 | Google `gemini-2.5-pro` | `auto`, `low`, `medium`, `high` | `medium` |
 | OpenAI reasoning models | `low`, `medium`, `high` | `medium` |
 

{kolega_code-0.8.4 → kolega_code-0.9.0}/kolega_code/__init__.py

@@ -151,4 +151,4 @@ __all__ = [
     "parse_git_status_output",
 ]
 
-__version__ = "0.8.4"
+__version__ = "0.9.0"

{kolega_code-0.8.4 → kolega_code-0.9.0}/kolega_code/agent/__init__.py

@@ -44,4 +44,4 @@ __all__ = [
     "ToolExtension",
 ]
 
-__version__ = "0.8.4"
+__version__ = "0.9.0"

{kolega_code-0.8.4 → kolega_code-0.9.0}/kolega_code/cli/app.py

@@ -714,6 +714,9 @@ class KolegaCodeApp(
         if not stripped_text or self.agent is None:
             if stripped_text:
                 self._set_settings_status(messages.SETTINGS_REQUIRED, tone="warning")
+                if self.config is None:
+                    self._set_composer_status(messages.DISCONNECTED_COMPOSER_PLACEHOLDER)
+                    self._show_composer_hint(messages.DISCONNECTED_ACTIVITY, tone="warning")
             return
         # Build attachments first (without clearing pending) so the vision gate
         # can block before we consume the composer text and pending images.
@@ -1340,6 +1343,10 @@ class KolegaCodeApp(
     def _set_chat_enabled(self, enabled: bool) -> None:
         composer = self.query_one("#composer", tui_widgets.ChatComposer)
         composer.disabled = not enabled or self._plan_decision_active or self._pending_approval is not None
+        if self.config is None and self.agent is None:
+            composer.placeholder = messages.DISCONNECTED_COMPOSER_PLACEHOLDER
+        elif enabled and composer.placeholder == messages.DISCONNECTED_COMPOSER_PLACEHOLDER:
+            composer.placeholder = messages.COMPOSER_PLACEHOLDER
 
     def _set_composer_status(self, status: str) -> None:
         self.query_one("#composer", tui_widgets.ChatComposer).placeholder = status
@@ -1460,10 +1467,19 @@ class KolegaCodeApp(
             if model
             else "not checked until a model is configured"
         )
-        return "\n".join(
+        startup_lines = [*tui_constants.STARTUP_WORDMARK, ""]
+        if self.config is None:
+            startup_lines.extend(
+                [
+                    messages.DISCONNECTED_HEADLINE,
+                    "",
+                    messages.DISCONNECTED_STARTUP_GUIDANCE,
+                    messages.DISCONNECTED_SIDEBAR_GUIDANCE,
+                    "",
+                ]
+            )
+        startup_lines.extend(
             [
-                *tui_constants.STARTUP_WORDMARK,
-                "",
                 f"Project: {self.project_path}",
                 f"Session: {session_id}",
                 f"Mode: {self.mode}",
@@ -1477,6 +1493,7 @@ class KolegaCodeApp(
                 f"Ctrl+C stop turn {theme.g(Glyph.BULLET_SEP)} Cmd+C copy selection {theme.g(Glyph.BULLET_SEP)} / commands",
             ]
         )
+        return "\n".join(startup_lines)
 
     def _startup_model(self) -> tuple[str, str]:
         if self.config is not None:

{kolega_code-0.8.4 → kolega_code-0.9.0}/kolega_code/cli/config.py

@@ -49,6 +49,7 @@ API_KEY_ENV = {
     ModelProvider.DEEPSEEK: "DEEPSEEK_API_KEY",
     ModelProvider.ZAI: "ZAI_API_KEY",
     ModelProvider.KIMI_CODING: "KIMI_CODING_API_KEY",
+    ModelProvider.OLLAMA_CLOUD: "OLLAMA_API_KEY",
 }
 
 DEFAULT_WEB_SEARCH_BACKEND = "duckduckgo"
@@ -399,6 +400,7 @@ def build_agent_config(
             deepseek_api_key=_api_key_for_provider(ModelProvider.DEEPSEEK, loaded_env, settings),
             zai_api_key=_api_key_for_provider(ModelProvider.ZAI, loaded_env, settings),
             kimi_coding_api_key=_api_key_for_provider(ModelProvider.KIMI_CODING, loaded_env, settings),
+            ollama_cloud_api_key=_api_key_for_provider(ModelProvider.OLLAMA_CLOUD, loaded_env, settings),
             environment=overrides.environment or loaded_env.get("KOLEGA_CODE_ENVIRONMENT", "development"),
             long_context_config=_model_config(long_provider, long_model, thinking_effort=active_thinking_effort),
             fast_config=_model_config(fast_provider, fast_model),

{kolega_code-0.8.4 → kolega_code-0.9.0}/kolega_code/cli/messages.py

@@ -9,6 +9,7 @@ from __future__ import annotations
 
 # Composer placeholders and modal prompts
 COMPOSER_PLACEHOLDER = "Ask Kolega Code..."
+DISCONNECTED_COMPOSER_PLACEHOLDER = "Connect a model in Settings before chatting."
 PLAN_READY_PLACEHOLDER = "Plan ready. Choose Implement plan or Discuss further."
 QUESTION_PLACEHOLDER = "Choose an option below or type a custom answer..."
 APPROVAL_PLACEHOLDER = "Choose whether to allow this action..."
@@ -18,6 +19,11 @@ THEME_PLACEHOLDER = "Choose a theme below or type a theme name..."
 
 # Durable transcript messages
 THREAD_RESET_MESSAGE = "Thread reset. Previous messages were cleared."
+DISCONNECTED_HEADLINE = "Not connected."
+DISCONNECTED_STARTUP_GUIDANCE = "Choose a provider and add an API key or sign in from the Settings tab before chatting."
+DISCONNECTED_SIDEBAR_GUIDANCE = "Press Ctrl+O to open the sidebar, then select Settings."
+DISCONNECTED_ACTIVITY = "Open Settings and connect a provider to start chatting."
+DISCONNECTED_MODEL = "not connected"
 TASK_LIST_EMPTY_MESSAGE = "No task list has been set."
 PLAN_EMPTY_MESSAGE = "No plan captured yet."
 

{kolega_code-0.8.4 → kolega_code-0.9.0}/kolega_code/cli/provider_registry.py

@@ -34,6 +34,7 @@ PROVIDER_LABELS: dict[ModelProvider, str] = {
     ModelProvider.FIREWORKS: "Fireworks",
     ModelProvider.TOGETHER: "Together AI",
     ModelProvider.DASHSCOPE: "DashScope / Qwen",
+    ModelProvider.OLLAMA_CLOUD: "Ollama Cloud",
 }
 
 # Friendly display names for models. Anything not listed falls back to its raw
@@ -84,6 +85,35 @@ MODEL_LABELS: dict[str, str] = {
     # DashScope / Qwen
     "qwen3-coder-plus": "Qwen3 Coder Plus",
     "qwen3-coder-flash": "Qwen3 Coder Flash",
+    # Ollama Cloud
+    "deepseek-v3.1:671b": "DeepSeek V3.1 671B",
+    "deepseek-v3.2": "DeepSeek V3.2",
+    "devstral-2:123b": "Devstral 2 123B",
+    "devstral-small-2:24b": "Devstral Small 2 24B",
+    "gemini-3-flash-preview": "Gemini 3 Flash Preview",
+    "gemma3:4b": "Gemma 3 4B",
+    "gemma3:12b": "Gemma 3 12B",
+    "gemma3:27b": "Gemma 3 27B",
+    "gemma4:31b": "Gemma 4 31B",
+    "glm-4.7": "GLM-4.7",
+    "glm-5": "GLM-5",
+    "gpt-oss:20b": "GPT-OSS 20B",
+    "gpt-oss:120b": "GPT-OSS 120B",
+    "kimi-k2.5": "Kimi K2.5",
+    "minimax-m2.1": "MiniMax M2.1",
+    "minimax-m2.5": "MiniMax M2.5",
+    "minimax-m2.7": "MiniMax M2.7",
+    "minimax-m3": "MiniMax M3",
+    "ministral-3:3b": "Ministral 3 3B",
+    "ministral-3:8b": "Ministral 3 8B",
+    "ministral-3:14b": "Ministral 3 14B",
+    "mistral-large-3:675b": "Mistral Large 3 675B",
+    "nemotron-3-nano:30b": "Nemotron 3 Nano 30B",
+    "nemotron-3-super": "Nemotron 3 Super",
+    "nemotron-3-ultra": "Nemotron 3 Ultra",
+    "qwen3-coder-next": "Qwen3 Coder Next",
+    "qwen3-coder:480b": "Qwen3 Coder 480B",
+    "qwen3.5:397b": "Qwen3.5 397B",
 }
 
 # Per-provider default model used when only a provider is selected. Covers the
@@ -101,6 +131,7 @@ PROVIDER_DEFAULT_MODEL: dict[ModelProvider, str] = {
     ModelProvider.FIREWORKS: "accounts/fireworks/models/glm-5p2",
     ModelProvider.TOGETHER: "moonshotai/Kimi-K2.7-Code",
     ModelProvider.DASHSCOPE: "qwen3-coder-plus",
+    ModelProvider.OLLAMA_CLOUD: "gpt-oss:20b",
 }
 
 UI_DEFAULT_PROVIDER = ModelProvider.MOONSHOT.value
@@ -130,6 +161,8 @@ def _api_key_env(provider: ModelProvider) -> str:
     """
     if provider == ModelProvider.OPENAI_CHATGPT:
         return ""
+    if provider == ModelProvider.OLLAMA_CLOUD:
+        return "OLLAMA_API_KEY"
     return f"{provider.value.upper()}_API_KEY"
 
 

{kolega_code-0.8.4 → kolega_code-0.9.0}/kolega_code/cli/tui/agent_runtime.py

@@ -4,8 +4,6 @@ from __future__ import annotations
 
 import asyncio
 
-from textual.widgets import TabbedContent
-
 from kolega_code.agent import AgentConfig, AgentEvent, CoderAgent, PlanningAgent, PromptExtension, ToolExtension
 from kolega_code.agent.prompt_provider import AgentMode
 from kolega_code.hooks import HookDispatcher, HookEvent, load_hook_config, project_hooks_present
@@ -198,7 +196,6 @@ class AgentRuntimeMixin:
             self._refresh_status_dashboard()
             self._set_settings_status(messages.SETTINGS_INCOMPLETE.format(error=exc), tone="error")
             self._ensure_startup_entry()
-            self.query_one("#events", TabbedContent).active = "settings_pane"
             return
 
         self.config = config

{kolega_code-0.8.4 → kolega_code-0.9.0}/kolega_code/cli/tui/status_dashboard.py

@@ -38,7 +38,12 @@ class StatusDashboardMixin:
 
     def _format_status_dashboard(self) -> str:
         state = self._status_state
-        provider_model = f"{state.provider}/{state.model}" if state.model else state.provider
+        disconnected = self.config is None
+        provider_model = (
+            messages.DISCONNECTED_MODEL
+            if disconnected
+            else (f"{state.provider}/{state.model}" if state.model else state.provider)
+        )
         effort = state.thinking_effort or "not supported"
         mode = state.mode.title()
         permission_mode = state.permission_mode.title()
@@ -83,7 +88,7 @@ class StatusDashboardMixin:
             f"{label('Context')}\n"
             f"{context_lines}\n\n"
             f"{label('Activity')}\n"
-            f"{escape(state.activity)}"
+            f"{escape(messages.DISCONNECTED_ACTIVITY if disconnected else state.activity)}"
         )
 
     def _context_bar(self, usage_percentage: float) -> str:

{kolega_code-0.8.4 → kolega_code-0.9.0}/kolega_code/config.py

@@ -23,6 +23,7 @@ class ModelProvider(str, Enum):
     DEEPSEEK = "deepseek"
     ZAI = "zai"
     KIMI_CODING = "kimi_coding"
+    OLLAMA_CLOUD = "ollama_cloud"
 
 
 class AgentRole(str, Enum):
@@ -131,6 +132,7 @@ class AgentConfig(BaseModel):
     deepseek_api_key: Optional[str] = Field(default=None, description="API key for DeepSeek")
     zai_api_key: Optional[str] = Field(default=None, description="API key for Z.AI (GLM Coding Plan)")
     kimi_coding_api_key: Optional[str] = Field(default=None, description="API key for Kimi Coding Plan")
+    ollama_cloud_api_key: Optional[str] = Field(default=None, description="API key for Ollama Cloud")
 
     # ChatGPT-subscription OAuth credentials (used instead of an api key for the
     # OPENAI_CHATGPT provider). The live, refreshing token manager is attached
@@ -221,6 +223,7 @@ class AgentConfig(BaseModel):
             ModelProvider.DEEPSEEK: self.deepseek_api_key,
             ModelProvider.ZAI: self.zai_api_key,
             ModelProvider.KIMI_CODING: self.kimi_coding_api_key,
+            ModelProvider.OLLAMA_CLOUD: self.ollama_cloud_api_key,
             ModelProvider.LLAMA: None,  # Local model, no API key needed
         }
         return api_key_map[provider]

{kolega_code-0.8.4 → kolega_code-0.9.0}/kolega_code/llm/client.py

@@ -135,6 +135,7 @@ class LLMClient:
                 "deepseek": AnthropicProvider,
                 "zai": AnthropicProvider,
                 "kimi_coding": AnthropicProvider,
+                "ollama_cloud": OpenAIProvider,
             }
 
             base_urls: Dict[str, str] = {
@@ -150,6 +151,7 @@ class LLMClient:
                 "deepseek": "https://api.deepseek.com/anthropic",
                 "zai": "https://api.z.ai/api/anthropic",
                 "kimi_coding": "https://api.kimi.com/coding",
+                "ollama_cloud": "https://ollama.com/v1",
             }
 
             provider_class = providers.get(provider.lower())

{kolega_code-0.8.4 → kolega_code-0.9.0}/kolega_code/llm/exceptions.py

@@ -141,6 +141,7 @@ def _provider_display_name(provider: str | None) -> str:
         ModelProvider.MOONSHOT.value: "Moonshot",
         ModelProvider.DEEPSEEK.value: "DeepSeek",
         ModelProvider.LLAMA.value: "Llama",
+        ModelProvider.OLLAMA_CLOUD.value: "Ollama Cloud",
     }
     if not provider:
         return "The selected provider"

{kolega_code-0.8.4 → kolega_code-0.9.0}/kolega_code/llm/models.py

@@ -934,8 +934,10 @@ class MessageChunk:
         delta = chunk.choices[0].delta
 
         # Fireworks and other OpenAI-compatible reasoning models stream their
-        # reasoning separately from answer text.
-        reasoning_content = getattr(delta, "reasoning_content", None)
+        # reasoning separately from answer text. Ollama Cloud uses the
+        # OpenAI-compatible non-standard "reasoning" field instead of
+        # "reasoning_content".
+        reasoning_content = getattr(delta, "reasoning_content", None) or getattr(delta, "reasoning", None)
         if reasoning_content:
             return cls(type="thinking", thinking=reasoning_content)
 
@@ -1142,8 +1144,9 @@ class Message:
         tool_use_blocks = []
 
         # OpenAI-compatible reasoning models (including Fireworks via Chat
-        # Completions) may return reasoning separately from answer text.
-        reasoning_content = getattr(message, "reasoning_content", None)
+        # Completions) may return reasoning separately from answer text. Ollama
+        # Cloud uses the OpenAI-compatible non-standard "reasoning" field.
+        reasoning_content = getattr(message, "reasoning_content", None) or getattr(message, "reasoning", None)
         if reasoning_content:
             content_blocks.append(ThinkingBlock(thinking=reasoning_content))
 

{kolega_code-0.8.4 → kolega_code-0.9.0}/kolega_code/llm/providers/openai.py

@@ -56,7 +56,9 @@ class OpenAIStreamWrapper:
                     if content:
                         self.final_content += content
 
-                    reasoning_content = getattr(delta, "reasoning_content", None) or ""
+                    reasoning_content = (
+                        getattr(delta, "reasoning_content", None) or getattr(delta, "reasoning", None) or ""
+                    )
                     if reasoning_content:
                         self.final_reasoning_content += reasoning_content