janito 2.19.1__tar.gz → 2.20.0__tar.gz

{janito-2.19.1 → janito-2.20.0}/CHANGELOG.md

@@ -2,6 +2,25 @@
 
 All notable changes to this project will be documented in this file.
 
+## [2.20.0] - 2025-08-10
+
+### Added
+
+- **Mistral AI Provider**: Added complete support for Mistral AI models
+  - **General Purpose Models**:
+    - `mistral-large-latest`: 128k context, most capable model (default)
+    - `mistral-medium-latest`: 32k context, balanced performance
+    - `mistral-small-latest`: 32k context, compact and efficient
+  - **Code-Focused Models**:
+    - `codestral-latest`: 256k context, specialized for code generation
+    - `codestral-2405`: Previous version of code-focused model
+  - **Development-Focused Models**:
+    - `devstral-small-latest`: 128k context, optimized for agentic tool use
+    - `devstral-medium-latest`: 128k context, enhanced agentic capabilities
+- **Mistral Setup Guide**: Comprehensive documentation for Mistral AI integration
+- **Tool Support**: Full tool/function calling support across all Mistral models
+- **API Integration**: Uses OpenAI-compatible API format via Mistral's La Plateforme
+
 ## [2.19.0] - 2025-08-08
 
 ### Added

{janito-2.19.1 → janito-2.20.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: janito
-Version: 2.19.1
+Version: 2.20.0
 Summary: A new Python package called janito.
 Author-email: João Pinto <janito@ikignosis.org>
 Project-URL: Homepage, https://github.com/ikignosis/janito

{janito-2.19.1 → janito-2.20.0}/docs/index.md

@@ -39,6 +39,7 @@ Janito supports multiple AI providers:
 - [OpenAI](openai-setup.md)
 - [DeepSeek](deepseek-setup.md)
 - [MoonshotAI](moonshotai-setup.md)
+- [Mistral](mistral-setup.md)
 - [Z.ai](z-ai-setup.md)
 - [Cerebras](cerebras-setup.md)
 

janito-2.20.0/docs/mistral-setup.md

@@ -0,0 +1,180 @@
+# Mistral Setup Guide
+
+This guide walks you through setting up Mistral AI models with Janito.
+
+## Overview
+
+Mistral AI provides a comprehensive suite of models including:
+- **General-purpose models**: Mistral Small, Medium, and Large
+- **Code-focused models**: Codestral for code generation and completion
+- **Development-focused models**: Devstral for agentic software development
+
+## Prerequisites
+
+1. **Python 3.8+** installed
+2. **Janito** installed (`pip install janito`)
+3. **Mistral API Key** from [La Plateforme](https://console.mistral.ai/)
+
+## Quick Setup
+
+### 1. Get Your API Key
+
+1. Visit [Mistral AI La Plateforme](https://console.mistral.ai/)
+2. Sign up or log in to your account
+3. Navigate to **API Keys** section
+4. Create a new API key
+5. Copy the key for use in Janito
+
+### 2. Configure Janito
+
+Set your Mistral API key using one of these methods:
+
+#### Method 1: Environment Variable (Recommended)
+
+```bash
+# Linux/macOS
+export MISTRAL_API_KEY="your-api-key-here"
+
+# Windows (PowerShell)
+$env:MISTRAL_API_KEY="your-api-key-here"
+
+# Windows (Command Prompt)
+set MISTRAL_API_KEY=your-api-key-here
+```
+
+#### Method 2: Janito CLI
+
+```bash
+janito set-api-key mistral your-api-key-here
+```
+
+#### Method 3: Interactive Setup
+
+```bash
+janito chat
+# Then use the !model command to select mistral and follow prompts
+```
+
+## Available Models
+
+### General Purpose Models
+- **mistral-large-latest** (default) - Most capable with 128k context
+- **mistral-medium-latest** - Balanced performance with 32k context
+- **mistral-small-latest** - Compact and efficient with 32k context
+
+### Code-Focused Models
+- **codestral-latest** - Specialized for code generation with 256k context
+- **codestral-2405** - Previous version of code-focused model
+
+### Development-Focused Models
+- **devstral-small-latest** - Optimized for agentic tool use
+- **devstral-medium-latest** - Enhanced agentic capabilities
+
+## Usage Examples
+
+### Basic Chat
+
+```bash
+# Start chat with default model (mistral-large-latest)
+janito chat --provider mistral
+
+# Use specific model
+janito chat --provider mistral --model codestral-latest
+```
+
+### Single-Shot Mode
+
+```bash
+# Quick question
+janito "Explain quantum computing" --provider mistral
+
+# Code generation
+janito "Write a Python function for fibonacci" --provider mistral --model codestral-latest
+```
+
+### Configuration File
+
+You can also configure Mistral in your Janito configuration:
+
+```yaml
+# ~/.janito/config.yaml
+providers:
+  mistral:
+    api_key: "your-api-key-here"
+    model: "mistral-large-latest"
+```
+
+## Model Selection Tips
+
+- **General conversations**: Use `mistral-large-latest` for best results
+- **Code generation**: Use `codestral-latest` for programming tasks
+- **Development workflows**: Use `devstral-*` models for agentic development
+- **Cost optimization**: Use `mistral-small-latest` for simpler tasks
+
+## Troubleshooting
+
+### Common Issues
+
+1. **"Invalid API key" error**
+   - Verify your API key is correct
+   - Check if the key has proper permissions
+   - Ensure no extra spaces in the key
+
+2. **"Model not found" error**
+   - Check available models with: `janito list-models --provider mistral`
+   - Ensure you're using the exact model name
+
+3. **Rate limiting**
+   - Mistral has rate limits based on your plan
+   - Consider upgrading your plan or implementing retry logic
+
+### Debug Mode
+
+Enable debug logging to see API requests:
+
+```bash
+# Enable HTTP debug for Mistral
+export MISTRAL_DEBUG_HTTP=1
+janito chat --provider mistral --verbose-api
+```
+
+### Testing Connection
+
+Test your setup:
+
+```bash
+# Ping the provider
+janito ping-providers
+
+# List available models
+janito list-models --provider mistral
+
+# Test with a simple query
+janito "Hello, Mistral!" --provider mistral
+```
+
+## Advanced Configuration
+
+### Custom Base URL
+
+If you need to use a different endpoint (e.g., for enterprise deployments):
+
+```bash
+janito set-config mistral.base_url "https://your-custom-endpoint.com/v1"
+```
+
+### Model Parameters
+
+You can set default parameters for Mistral models:
+
+```bash
+janito set-config mistral.temperature 0.5
+janito set-config mistral.max_tokens 2000
+```
+
+## Support
+
+For additional help:
+- Check the [Mistral AI documentation](https://docs.mistral.ai/)
+- Review [Janito troubleshooting guides](guides/configuration.md)
+- Join our [Discord community](https://discord.gg/janito)

janito-2.20.0/docs/provider-platform-access.md

@@ -0,0 +1,26 @@
+# Provider Platform and Documentation Access
+
+This table shows the accessibility status of each provider's platform/dashboard and API documentation.
+
+| Provider | Platform Access | Documentation | API Docs | Status |
+|----------|-----------------|---------------|----------|--------|
+| **Alibaba** | [Alibaba Cloud Console](https://account.alibabacloud.com) | [Model Studio Help](https://www.alibabacloud.com/help/en/model-studio) | [API Reference](https://www.alibabacloud.com/help/en/model-studio/api-reference) | ✅ Full Access |
+| **Anthropic** | [Console](https://console.anthropic.com) | [API Docs](https://docs.anthropic.com/en/api/getting-started) | [API Reference](https://docs.anthropic.com/en/api/messages) | ⚠️ Console Blocked |
+| **Cerebras** | [API Dashboard](https://api.cerebras.ai) | [Inference Docs](https://cerebras.ai/inference) | [API Docs](https://cerebras.ai/inference) | ✅ Full Access |
+| **DeepSeek** | [Platform](https://platform.deepseek.com) | [API Docs](https://platform.deepseek.com/api-docs) | [API Reference](https://platform.deepseek.com/api-docs) | ❌ All Blocked |
+| **Google** | [AI Studio](https://aistudio.google.com) | [Gemini API Docs](https://ai.google.dev/gemini-api/docs) | [API Reference](https://ai.google.dev/gemini-api/docs) | ✅ Full Access |
+| **MoonshotAI** | [Platform](https://platform.moonshot.ai) | [API Docs](https://platform.moonshot.ai/docs) | [API Reference](https://platform.moonshot.ai/docs) | ✅ Full Access |
+| **OpenAI** | [Platform](https://platform.openai.com) | [API Docs](https://platform.openai.com/docs) | [API Reference](https://platform.openai.com/docs/api-reference) | ❌ All Blocked |
+| **Z.ai** | [API Management](https://z.ai/manage-apikey/apikey-list) | [Model API Docs](https://z.ai/model-api) | [API Reference](https://z.ai/model-api) | ✅ Full Access |
+
+## Status Legend
+
+- ✅ **Full Access**: Both platform and documentation are publicly accessible
+- ⚠️ **Partial Access**: Documentation available but platform/dashboard blocked
+- ❌ **Blocked**: Both platform and documentation are inaccessible (403 errors)
+
+## Notes
+
+- **API Endpoints**: All API base URLs return 404 when accessed directly without authentication (expected behavior)
+- **Authentication**: All providers require valid API keys for actual API usage
+- **Regional Restrictions**: Some providers may have additional access restrictions based on geographic location

{janito-2.19.1 → janito-2.20.0}/docs/supported-providers-models.md

@@ -2,6 +2,8 @@
 
 This page lists the supported providers and their available models, organized by openness and sorted alphabetically within each category.
 
+> 📋 **Platform & Documentation Access**: For information about which providers have publicly accessible platforms and documentation vs. blocked access, see [Platform & Documentation Access](provider-platform-access.md).
+
 ## Open-Source / Open-Weight Models
 
 ### Alibaba
@@ -40,6 +42,20 @@ This page lists the supported providers and their available models, organized by
 - **DeepSeek Chat** (default) - General purpose chat model
 - **DeepSeek Reasoner** - Specialized for complex reasoning tasks
 
+### Mistral
+
+- **Mistral Large Latest** (default) - Most capable Mistral model with 128k context
+- **Mistral Medium Latest** - Balanced performance with 32k context
+- **Mistral Small Latest** - Compact and efficient with 32k context
+- **Codestral Latest** - Specialized for code generation with 256k context
+- **Codestral 2405** - Previous version of code-focused model
+- **Devstral Small Latest** - Optimized for agentic tool use in software development
+- **Devstral Medium Latest** - Enhanced agentic capabilities for development tasks
+
+Mistral provides both general-purpose and specialized models, with Codestral specifically designed for code generation and Devstral for agentic software development.
+
+For setup instructions, see the [Mistral Setup Guide](mistral-setup.md).
+
 ### MoonshotAI
 
 - **Kimi K2 Turbo Preview** (default) - Latest generation with enhanced performance

{janito-2.19.1 → janito-2.20.0}/janito/cli/cli_commands/list_models.py

@@ -2,8 +2,8 @@
 CLI Command: List models for the specified/current provider
 """
 
-from janito.cli.cli_commands.model_utils import _print_models_table
 import sys
+from janito.cli.cli_commands.model_utils import _print_models_table
 
 _provider_instance = None
 
@@ -36,8 +36,6 @@ def handle_list_models(args, provider_instance):
             
             for m in models:
                 table.add_row(str(m))
-                
-            import sys
             if sys.stdout.isatty():
                 shared_console.print(table)
             else:

{janito-2.19.1 → janito-2.20.0}/janito/drivers/openai/driver.py

@@ -64,14 +64,20 @@ class OpenAIModelDriver(LLMDriver):
         if config.model:
             api_kwargs["model"] = config.model
         # Prefer max_completion_tokens if present, else fallback to max_tokens (for backward compatibility)
-        if (
-            hasattr(config, "max_completion_tokens")
-            and config.max_completion_tokens is not None
-        ):
-            api_kwargs["max_completion_tokens"] = int(config.max_completion_tokens)
+        # Skip max_completion_tokens for Mistral as their API doesn't support it
+        is_mistral = config.base_url and "mistral.ai" in str(config.base_url)
+        if not is_mistral:
+            if (
+                hasattr(config, "max_completion_tokens")
+                and config.max_completion_tokens is not None
+            ):
+                api_kwargs["max_completion_tokens"] = int(config.max_completion_tokens)
+            elif hasattr(config, "max_tokens") and config.max_tokens is not None:
+                # For models that do not support 'max_tokens', map to 'max_completion_tokens'
+                api_kwargs["max_completion_tokens"] = int(config.max_tokens)
         elif hasattr(config, "max_tokens") and config.max_tokens is not None:
-            # For models that do not support 'max_tokens', map to 'max_completion_tokens'
-            api_kwargs["max_completion_tokens"] = int(config.max_tokens)
+            # For Mistral, use max_tokens directly
+            api_kwargs["max_tokens"] = int(config.max_tokens)
         for p in (
             "temperature",
             "top_p",

{janito-2.19.1 → janito-2.20.0}/janito/providers/__init__.py

@@ -8,3 +8,4 @@ import janito.providers.moonshotai.provider
 import janito.providers.alibaba.provider
 import janito.providers.zai.provider
 import janito.providers.cerebras.provider
+import janito.providers.mistral.provider

janito-2.20.0/janito/providers/mistral/__init__.py

@@ -0,0 +1 @@
+# Codestral provider module

janito-2.20.0/janito/providers/mistral/model_info.py

@@ -0,0 +1,81 @@
+from janito.llm.model import LLMModelInfo
+
+MODEL_SPECS = {
+    "codestral-latest": LLMModelInfo(
+        name="codestral-latest",
+        context=256000,
+        max_input=250000,
+        max_cot="N/A",
+        max_response=4096,
+        thinking_supported=False,
+        default_temp=0.2,
+        open="mistral",
+        driver="OpenAIModelDriver",
+    ),
+    "codestral-2405": LLMModelInfo(
+        name="codestral-2405",
+        context=256000,
+        max_input=250000,
+        max_cot="N/A",
+        max_response=4096,
+        thinking_supported=False,
+        default_temp=0.2,
+        open="mistral",
+        driver="OpenAIModelDriver",
+    ),
+    "mistral-small-latest": LLMModelInfo(
+        name="mistral-small-latest",
+        context=32000,
+        max_input=28000,
+        max_cot="N/A",
+        max_response=4096,
+        thinking_supported=False,
+        default_temp=0.7,
+        open="mistral",
+        driver="OpenAIModelDriver",
+    ),
+    "mistral-medium-latest": LLMModelInfo(
+        name="mistral-medium-latest",
+        context=32000,
+        max_input=28000,
+        max_cot="N/A",
+        max_response=4096,
+        thinking_supported=False,
+        default_temp=0.7,
+        open="mistral",
+        driver="OpenAIModelDriver",
+    ),
+    "mistral-large-latest": LLMModelInfo(
+        name="mistral-large-latest",
+        context=128000,
+        max_input=120000,
+        max_cot="N/A",
+        max_response=4096,
+        thinking_supported=False,
+        default_temp=0.7,
+        open="mistral",
+        driver="OpenAIModelDriver",
+    ),
+    "devstral-small-latest": LLMModelInfo(
+        name="devstral-small-latest",
+        context=128000,
+        max_input=120000,
+        max_cot="N/A",
+        max_response=4096,
+        thinking_supported=False,
+        default_temp=0.7,
+        open="mistral",
+        driver="OpenAIModelDriver",
+    ),
+    "devstral-medium-latest": LLMModelInfo(
+        name="devstral-medium-latest",
+        context=128000,
+        max_input=120000,
+        max_cot="N/A",
+        max_response=4096,
+        thinking_supported=False,
+        default_temp=0.7,
+        open="mistral",
+        driver="OpenAIModelDriver",
+    ),
+}

janito-2.20.0/janito/providers/mistral/provider.py

@@ -0,0 +1,124 @@
+from janito.llm.provider import LLMProvider
+from janito.llm.model import LLMModelInfo
+from janito.llm.auth import LLMAuthManager
+from janito.llm.driver_config import LLMDriverConfig
+from janito.drivers.openai.driver import OpenAIModelDriver
+from janito.tools import get_local_tools_adapter
+from janito.providers.registry import LLMProviderRegistry
+from .model_info import MODEL_SPECS
+
+available = OpenAIModelDriver.available
+unavailable_reason = OpenAIModelDriver.unavailable_reason
+
+
+class MistralProvider(LLMProvider):
+    name = "mistral"
+    NAME = "mistral"
+    MAINTAINER = "João Pinto <janito@ikignosis.org>"
+    MODEL_SPECS = MODEL_SPECS
+    DEFAULT_MODEL = "mistral-large-latest"
+
+    def __init__(
+        self, auth_manager: LLMAuthManager = None, config: LLMDriverConfig = None
+    ):
+        self._tools_adapter = get_local_tools_adapter()
+        self._driver = None
+
+        if not self.available:
+            return
+
+        self._initialize_config(auth_manager, config)
+        self._setup_model_config()
+
+    def _initialize_config(self, auth_manager, config):
+        """Initialize configuration and API key."""
+        self.auth_manager = auth_manager or LLMAuthManager()
+        self._api_key = self.auth_manager.get_credentials(type(self).NAME)
+        if not self._api_key:
+            from janito.llm.auth_utils import handle_missing_api_key
+
+            handle_missing_api_key(self.name, "MISTRAL_API_KEY")
+
+        self._driver_config = config or LLMDriverConfig(model=None)
+        if not self._driver_config.model:
+            self._driver_config.model = self.DEFAULT_MODEL
+        if not self._driver_config.api_key:
+            self._driver_config.api_key = self._api_key
+
+        # Set Mistral-specific base URL
+        self._driver_config.base_url = "https://api.mistral.ai/v1"
+
+    def _setup_model_config(self):
+        """Configure token limits based on model specifications."""
+        model_name = self._driver_config.model
+        model_spec = self.MODEL_SPECS.get(model_name)
+
+        # Reset token parameters
+        if hasattr(self._driver_config, "max_tokens"):
+            self._driver_config.max_tokens = None
+        if hasattr(self._driver_config, "max_completion_tokens"):
+            self._driver_config.max_completion_tokens = None
+
+        if model_spec:
+            if getattr(model_spec, "thinking_supported", False):
+                max_cot = getattr(model_spec, "max_cot", None)
+                if max_cot and max_cot != "N/A":
+                    self._driver_config.max_tokens = int(max_cot)
+            else:
+                max_response = getattr(model_spec, "max_response", None)
+                if max_response and max_response != "N/A":
+                    self._driver_config.max_tokens = int(max_response)
+
+        self.fill_missing_device_info(self._driver_config)
+
+    @property
+    def driver(self) -> OpenAIModelDriver:
+        if not self.available:
+            raise ImportError(f"MistralProvider unavailable: {self.unavailable_reason}")
+        return self._driver
+
+    @property
+    def available(self):
+        return available
+
+    @property
+    def unavailable_reason(self):
+        return unavailable_reason
+
+    def create_driver(self):
+        """
+        Creates and returns a new OpenAIModelDriver instance with input/output queues.
+        """
+        driver = OpenAIModelDriver(
+            tools_adapter=self._tools_adapter, provider_name=self.NAME
+        )
+        driver.config = self._driver_config
+        # NOTE: The caller is responsible for calling driver.start() if background processing is needed.
+        return driver
+
+    def create_agent(self, tools_adapter=None, agent_name: str = None, **kwargs):
+        from janito.llm.agent import LLMAgent
+
+        # Always create a new driver with the passed-in tools_adapter
+        if tools_adapter is None:
+            tools_adapter = get_local_tools_adapter()
+        # Should use new-style driver construction via queues/factory (handled elsewhere)
+        raise NotImplementedError(
+            "create_agent must be constructed via new factory using input/output queues and config."
+        )
+
+    @property
+    def model_name(self):
+        return self._driver_config.model
+
+    @property
+    def driver_config(self):
+        """Public, read-only access to the provider's LLMDriverConfig object."""
+        return self._driver_config
+
+    def execute_tool(self, tool_name: str, event_bus, *args, **kwargs):
+        self._tools_adapter.event_bus = event_bus
+        return self._tools_adapter.execute_by_name(tool_name, *args, **kwargs)
+
+
+LLMProviderRegistry.register(MistralProvider.NAME, MistralProvider)

{janito-2.19.1 → janito-2.20.0}/janito.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: janito
-Version: 2.19.1
+Version: 2.20.0
 Summary: A new Python package called janito.
 Author-email: João Pinto <janito@ikignosis.org>
 Project-URL: Homepage, https://github.com/ikignosis/janito

{janito-2.19.1 → janito-2.20.0}/janito.egg-info/SOURCES.txt

@@ -35,8 +35,10 @@ docs/event-bus.md
 docs/index.md
 docs/llm-drivers-required-config.md
 docs/llm-drivers.md
+docs/mistral-setup.md
 docs/moonshotai-setup.md
 docs/openai-setup.md
+docs/provider-platform-access.md
 docs/security.md
 docs/supported-providers-models.md
 docs/tools-index.md
@@ -236,6 +238,9 @@ janito/providers/deepseek/provider.py
 janito/providers/google/__init__.py
 janito/providers/google/model_info.py
 janito/providers/google/provider.py
+janito/providers/mistral/__init__.py
+janito/providers/mistral/model_info.py
+janito/providers/mistral/provider.py
 janito/providers/moonshotai/__init__.py
 janito/providers/moonshotai/model_info.py
 janito/providers/moonshotai/provider.py

{janito-2.19.1 → janito-2.20.0}/mkdocs.yml

@@ -35,6 +35,7 @@ nav:
       - API Reference: reference/api.md
   - Supported Providers & Models:
       - Overview: supported-providers-models.md
+      - Platform & Documentation Access: provider-platform-access.md
       - OpenAI Setup: openai-setup.md
       - DeepSeek Setup: deepseek-setup.md
       - MoonshotAI Setup: moonshotai-setup.md