feat: Release v1.8.0 - Context Management and Dependency Refactor

- Context: Added per-provider 'context_limit' and automatic history pruning/distillation to prevent failover loops.
- Config: Added '/config budget' command and '/config set ... context_limit' support.
- Fix: Resolved critical code duplication in OpenAICompatibleProvider causing double responses.
- Fix: Fixed UI buffer duplication during retry/failover events.
- Refactor: Consolidated dependencies into pyproject.toml, simplifying requirements.txt and CI/CD.
- Tests: Converted verification scripts to pytest and achieved 100% pass rate.
- Docs: Updated documentation for context management and new commands.

Author: Pomilon

.github/workflows/ci.yml

@@ -20,12 +20,11 @@ jobs:
     - name: Install dependencies
       run: |
         python -m pip install --upgrade pip
-        pip install -e .
-        pip install pytest pytest-asyncio beautifulsoup4 requests pydantic-settings
+        pip install -e ".[dev]"
         
-    - name: Run unit tests
+    - name: Run all tests
       run: |
-        pytest tests/test_components.py tests/test_v1_2_features.py tests/test_web_v1_3.py
+        pytest tests/
         
     - name: Set up Docker
       uses: docker-practice/actions-setup-docker@master

README.md

@@ -1,15 +1,15 @@
-# Plexir 🌌
+# Plexir
 
 **Plexir** is a modular, secure, and beautiful AI-powered terminal workspace. Designed for developers who live in the CLI, it combines multi-provider LLM orchestration, advanced agentic tools, and a secure, persistent Docker sandbox into a single keyboard-centric interface.
 
 ![Plexir UI](assets/image.png)
 
-[![Version](https://img.shields.io/badge/version-1.7.0-blue.svg)](https://github.com/pomilon/plexir)
+[![Version](https://img.shields.io/badge/version-1.8.0-blue.svg)](https://github.com/pomilon/plexir)
 [![License: MIT](https://img.shields.io/badge/License-MIT-yellow.svg)](https://opensource.org/licenses/MIT)
 
 ---
 
-## 🚀 Features
+## Features
 
 - **Multi-Provider Failover**: Seamlessly switch between Gemini, Groq, Cerebras, and OpenAI-compatible APIs. If one model hits a quota, Plexir automatically fails over to the next in your priority list.
 - **Economics & Metrics**: Real-time **Token Tracking** and **Cost Estimation** in the sidebar. Set a session budget via `/config budget` to prevent runaway costs.
@@ -18,6 +18,7 @@
     - **Persistent Memory Bank**: Semantic storage (`chromadb`) for long-term facts using `/memory save`.
     - **Rolling Summarization**: Automatically condenses long histories.
     - **Message Pinning**: `/session pin` ensures critical context is never lost.
+    - **Context Window Management**: Intelligent token counting and history pruning ensures conversations never exceed model limits, preventing failover loops.
 - **Persistent Docker Sandbox**: Launch with `--sandbox` to give the AI its own persistent Linux "computer." All tools (file system, git, shell) are automatically redirected inside the container.
 - **Deep MCP Integration**: Fully supports **Model Context Protocol (MCP)**, including dynamic discovery of tools, **Resources**, **Resource Templates**, and **Prompts** from MCP servers.
 - **Smart Agent Capabilities**:
@@ -42,7 +43,7 @@
 
 ---
 
-## 📦 Installation
+## Installation
 
 ### Prerequisites
 - Python 3.10 or higher.
@@ -61,7 +62,7 @@
 
 ---
 
-## 🛠 Usage
+## Usage
 
 Start Plexir from any terminal:
 ```bash
@@ -90,7 +91,7 @@ plexir --sandbox
 
 ---
 
-## ⚙️ Configuration
+## Configuration
 
 Plexir stores its configuration in `~/.plexir/config.json`. You can manage this file manually or using the in-app `/config` commands. 
 
@@ -101,7 +102,7 @@ Example failover hierarchy:
 
 ---
 
-## 📄 Documentation
+## Documentation
 
 Detailed guides are available in the `docs/` directory:
 - [Getting Started](docs/getting_started.md)
@@ -113,14 +114,14 @@ Detailed guides are available in the `docs/` directory:
 
 ---
 
-## 🤝 Contributing
+## Contributing
 
 Contributions are welcome! Please see our [Contributing Guidelines](CONTRIBUTING.md) and [Code of Conduct](CODE_OF_CONDUCT.md) for more details.
 
-## 🛡️ Security
+## Security
 
 If you discover a security vulnerability, please see our [Security Policy](SECURITY.md).
 
-## ⚖️ License
+## License
 
-Plexir is released under the [MIT License](LICENSE).
+Plexir is released under the [MIT License](LICENSE).

docs/COMMANDS.md

@@ -4,7 +4,7 @@ This document provides a comprehensive guide to all available slash commands in
 
 ---
 
-## 🚀 General Commands
+## General Commands
 
 ### `/help`
 Displays a summary of all primary slash commands.
@@ -23,7 +23,7 @@ Exits the Plexir application cleanly, stopping the sandbox if it's running.
 
 ---
 
-## ⚙️ Configuration (`/config`)
+## Configuration (`/config`)
 
 Manages providers, tool settings, and application preferences.
 
@@ -33,11 +33,16 @@ Shows the current provider failover order, tool configurations, and other app se
 #### `/config set <provider_name> <key> <value>`
 Updates a specific property for a provider.
 - **`<provider_name>`**: The name of the provider (e.g., `"Gemini Primary"`). Use quotes if it contains spaces.
-- **`<key>`**: The property to change (`api_key`, `model_name`, `type`, `base_url`).
+- **`<key>`**: The property to change (`api_key`, `model_name`, `type`, `base_url`, `context_limit`).
 - **`<value>`**: The new value.
-- **Example**: `/config set "Gemini Primary" api_key ghp_...`
+- **Example**: `/config set "Gemini Primary" context_limit 50000`
 - **Example**: `/config set "Groq Backup" model_name llama3-70b-8192`
 
+#### `/config budget <amount>`
+Sets a maximum dollar amount for the current session to prevent runaway costs.
+- **Example**: `/config budget 0.50`
+- **Example**: `/config budget 0` (disables limit)
+
 #### `/config tool <domain> <key> <value>`
 Sets a configuration value for a specific tool suite. This is how you provide tokens for external services.
 - **`<domain>`**: The tool domain (e.g., `git`, `github`).
@@ -67,7 +72,7 @@ Changes the failover priority of a provider.
 
 ---
 
-## 💾 Session Management (`/session`)
+## Session Management (`/session`)
 
 Saves and loads chat histories.
 
@@ -88,7 +93,7 @@ Deletes a saved session file.
 
 ---
 
-## 📹 Macro Management (`/macro`)
+## Macro Management (`/macro`)
 
 Records and plays back sequences of commands and prompts.
 
@@ -108,4 +113,4 @@ Lists all saved macros.
 
 #### `/macro delete <name>`
 Deletes a saved macro.
-- **Example**: `/macro delete old-macro`
+- **Example**: `/macro delete old-macro`

docs/configuration.md

@@ -26,6 +26,27 @@ Instead of plain text, you can use:
 - `env:VARIABLE_NAME`: Read from environment variables.
 - `keyring:username`: Read from the system keyring (service: `plexir`).
 
+## Context Management
+
+Plexir automatically manages the context window to prevent model errors when conversations get too long.
+
+### Automatic Limits
+Plexir comes with pre-configured token limits for popular models (e.g., 2M tokens for Gemini 1.5 Pro, 128k for GPT-4o). When the conversation history exceeds this limit, Plexir will:
+1.  **Preserve** the most recent messages.
+2.  **Preserve** system instructions.
+3.  **Summarize/Distill** the older parts of the conversation to save space while retaining context.
+
+### Manual Configuration (`context_limit`)
+You can override the default limit for any provider. This is useful for:
+- Testing how models behave with shorter context.
+- Forcing stricter limits on "Preview" models to save costs.
+
+To set a strict 50,000 token limit on a provider:
+```bash
+/config set "Gemini Primary" context_limit 50000
+```
+*Set to `0` or `null` to use the model's default.*
+
 ## Failover & Retries
 
 Plexir manages providers using a priority order defined in your config.
@@ -87,4 +108,4 @@ To prevent unexpected costs during long sessions, you can set a maximum dollar a
 *Set to `0` to disable the limit.*
 
 ### View current usage
-Usage metrics (Tokens and Estimated Cost) are always visible in the **System Status** sidebar.
+Usage metrics (Tokens and Estimated Cost) are always visible in the **System Status** sidebar.

docs/memory.md

@@ -2,7 +2,7 @@
 
 Plexir uses advanced techniques to manage long-running conversations, ensuring the model remains coherent even as the history grows.
 
-## 🧠 Persistent Memory Bank (New in v1.7)
+## Persistent Memory Bank (New in v1.7)
 
 Plexir now includes a **Long-Term Memory** system powered by `chromadb`. This allows the agent to store and recall specific facts across different sessions.
 
@@ -20,15 +20,15 @@ You can prompt the agent to remember things directly:
 Or ask it to recall:
 > "Where did I say my keys were?"
 
-## 🔄 Rolling Summarization
+## Rolling Summarization
 
 When a conversation history becomes too large (exceeding 40 messages), Plexir automatically triggers **Rolling Summarization**. 
 
 1. **How it works**: Plexir identifies older, unpinned messages and uses the primary LLM to condense them into a concise "BACKGROUND SUMMARY."
 2. **Context Preservation**: This summary is injected at the start of the conversation, allowing the model to remember high-level decisions and context while clearing out detailed token-heavy noise.
 3. **Automatic**: This happens in the background without user intervention.
 
-## 📌 Message Pinning
+## Message Pinning
 
 You can manually protect critical messages from being summarized or pruned using **Context Pinning**.
 
@@ -44,6 +44,6 @@ Use the `/session pin` command followed by the message number (visible in the hi
 /session unpin 5
 ```
 
-## 🧠 Distillation (Failover)
+## Distillation (Failover)
 
-During a **Provider Failover** (e.g., Gemini Primary hitting a quota), Plexir uses a "Distillation" process to transfer only the most essential recent context to the backup provider. This ensures a smooth transition with minimal latency and token waste.
+During a **Provider Failover** (e.g., Gemini Primary hitting a quota), Plexir uses a "Distillation" process to transfer only the most essential recent context to the backup provider. This ensures a smooth transition with minimal latency and token waste.

docs/sandbox.md

@@ -31,4 +31,4 @@ If you encounter issues with the sandbox:
    ```bash
    docker stop plexir-persistent-sandbox
    docker rm plexir-persistent-sandbox
-   ```
+   ```

docs/tools.md

@@ -65,4 +65,4 @@ Executes Python code in an isolated Docker container. This is useful for perform
 ## Critical Actions & HITL
 Tools marked as **Critical** in the table above will trigger a **Human-in-the-Loop (HITL)** confirmation dialog in the TUI. 
 *   **Visual Diffs:** For `write_file` and `edit_file`, the confirmation dialog displays a color-coded diff of the proposed changes.
-*   You must manually click **Confirm** before the action is executed.
+*   You must manually click **Confirm** before the action is executed.

plexir/__init__.py

@@ -1 +1 @@
-__version__ = "1.7.0"
+__version__ = "1.8.0"

plexir/core/config_manager.py

@@ -60,6 +60,7 @@ class ProviderConfig(BaseModel):
         "auto", 
         description="Authentication mode: 'auto', 'api_key', or 'oauth' (Gemini only)."
     )
+    context_limit: Optional[int] = Field(None, description="Max input tokens. If None, uses model default.")
 
     def get_api_key(self) -> Optional[str]:
         """Resolves the API key securely."""
@@ -121,6 +122,34 @@ class AppConfig(BaseModel):
         },
         description="Pricing map: model -> (prompt_price, completion_price) per 1M tokens."
     )
+    model_context_windows: Dict[str, int] = Field(
+        default_factory=lambda: {
+            # --- Google Gemini Series ---
+            "gemini-3-pro-preview": 2000000,
+            "gemini-3-flash-preview": 1000000,
+            "gemini-2.5-pro": 2000000,
+            "gemini-2.5-flash": 1000000,
+            "gemini-2.5-flash-lite": 1000000,
+            
+            # --- Anthropic Claude Series ---
+            "claude-4.5-sonnet": 200000,
+            "claude-4.5-haiku": 200000,
+            "claude-4-opus": 200000,
+
+            # --- OpenAI Series ---
+            "gpt-4o": 128000,
+            "gpt-4o-mini": 128000,
+            "gpt-oss-120b": 8192,
+            
+            # --- Specialized ---
+            "deepseek-v3": 64000,
+            "deepseek-reasoner": 64000,
+            "llama-3.3-70b-versatile": 128000,
+            "llama3.1-8b": 8192,
+            "llama3.1-70b": 8192,
+        },
+        description="Default context window sizes (in tokens) for known models."
+    )
     macros: Dict[str, List[str]] = Field(default_factory=dict)
     tool_configs: Dict[str, Dict[str, str]] = Field(default_factory=dict)