quantalogic 0.2.17__py3-none-any.whl → 0.2.18__py3-none-any.whl

quantalogic/agent.py

@@ -71,6 +71,8 @@ class Agent(BaseModel):
     max_output_tokens: int = DEFAULT_MAX_OUTPUT_TOKENS
     max_iterations: int = 30
     system_prompt: str = ""
+    compact_every_n_iterations: int | None = None  # Add this to the class attributes
+    max_tokens_working_memory: int | None = None  # Add max_tokens_working_memory attribute
 
     def __init__(
         self,
@@ -81,6 +83,8 @@ class Agent(BaseModel):
         task_to_solve: str = "",
         specific_expertise: str = "General AI assistant with coding and problem-solving capabilities",
         get_environment: Callable[[], str] = get_environment,
+        compact_every_n_iterations: int | None = None,  # New parameter
+        max_tokens_working_memory: int | None = None,  # New parameter to set max working memory tokens
     ):
         """Initialize the agent with model, memory, tools, and configurations."""
         try:
@@ -121,6 +125,15 @@ class Agent(BaseModel):
                 specific_expertise=specific_expertise,
                 event_emitter=event_emitter,
             )
+            
+            # Set the new compact_every_n_iterations parameter
+            self.compact_every_n_iterations = compact_every_n_iterations or self.max_iterations
+            logger.debug(f"Memory will be compacted every {self.compact_every_n_iterations} iterations")
+            
+            # Set the max_tokens_working_memory parameter
+            self.max_tokens_working_memory = max_tokens_working_memory
+            logger.debug(f"Max tokens for working memory set to: {self.max_tokens_working_memory}")
+            
             logger.debug("Agent initialized successfully.")
         except Exception as e:
             logger.error(f"Failed to initialize agent: {str(e)}")
@@ -265,9 +278,31 @@ class Agent(BaseModel):
         self.total_tokens = self.model.token_counter_with_history(message_history, prompt)
 
     def _compact_memory_if_needed(self, current_prompt: str = ""):
-        """Compacts the memory if it exceeds the maximum occupancy."""
+        """Compacts the memory if it exceeds the maximum occupancy or token limit."""
         ratio_occupied = self._calculate_context_occupancy()
-        if ratio_occupied >= MAX_OCCUPANCY:
+        
+        # Compact memory if any of these conditions are met:
+        # 1. Memory occupancy exceeds MAX_OCCUPANCY, or
+        # 2. Current iteration is a multiple of compact_every_n_iterations, or
+        # 3. Working memory exceeds max_tokens_working_memory (if set)
+        should_compact_by_occupancy = ratio_occupied >= MAX_OCCUPANCY
+        should_compact_by_iteration = (
+            self.compact_every_n_iterations is not None and 
+            self.current_iteration > 0 and 
+            self.current_iteration % self.compact_every_n_iterations == 0
+        )
+        should_compact_by_token_limit = (
+            self.max_tokens_working_memory is not None and 
+            self.total_tokens > self.max_tokens_working_memory
+        )
+        
+        if should_compact_by_occupancy or should_compact_by_iteration or should_compact_by_token_limit:
+            if should_compact_by_occupancy:
+                logger.debug(f"Memory compaction triggered: Occupancy {ratio_occupied}% exceeds {MAX_OCCUPANCY}%")
+            
+            if should_compact_by_iteration:
+                logger.debug(f"Memory compaction triggered: Iteration {self.current_iteration} is a multiple of {self.compact_every_n_iterations}")
+            
             self._emit_event("memory_full")
             self.memory.compact()
             self.total_tokens = self.model.token_counter_with_history(self.memory.memory, current_prompt)

quantalogic/agent_config.py

@@ -32,13 +32,21 @@ from quantalogic.tools import (
 MODEL_NAME = "deepseek/deepseek-chat"
 
 
-def create_agent(model_name: str, vision_model_name: str | None, no_stream: bool = False) -> Agent:
+def create_agent(
+    model_name: str, 
+    vision_model_name: str | None, 
+    no_stream: bool = False, 
+    compact_every_n_iteration: int | None = None,
+    max_tokens_working_memory: int | None = None
+) -> Agent:
     """Create an agent with the specified model and tools.
 
     Args:
         model_name (str): Name of the model to use
         vision_model_name (str | None): Name of the vision model to use
         no_stream (bool, optional): If True, the agent will not stream results.
+        compact_every_n_iteration (int | None, optional): Frequency of memory compaction.
+        max_tokens_working_memory (int | None, optional): Maximum tokens for working memory.
 
     Returns:
         Agent: An agent with the specified model and tools
@@ -66,16 +74,26 @@ def create_agent(model_name: str, vision_model_name: str | None, no_stream: bool
     return Agent(
         model_name=model_name,
         tools=tools,
+        compact_every_n_iterations=compact_every_n_iteration,
+        max_tokens_working_memory=max_tokens_working_memory,
     )
 
 
-def create_interpreter_agent(model_name: str, vision_model_name: str | None, no_stream: bool = False) -> Agent:
+def create_interpreter_agent(
+    model_name: str, 
+    vision_model_name: str | None, 
+    no_stream: bool = False, 
+    compact_every_n_iteration: int | None = None,
+    max_tokens_working_memory: int | None = None
+) -> Agent:
     """Create an interpreter agent with the specified model and tools.
 
     Args:
         model_name (str): Name of the model to use
         vision_model_name (str | None): Name of the vision model to use
         no_stream (bool, optional): If True, the agent will not stream results.
+        compact_every_n_iteration (int | None, optional): Frequency of memory compaction.
+        max_tokens_working_memory (int | None, optional): Maximum tokens for working memory.
 
     Returns:
         Agent: An interpreter agent with the specified model and tools
@@ -98,16 +116,29 @@ def create_interpreter_agent(model_name: str, vision_model_name: str | None, no_
         LLMTool(model_name=model_name, on_token=console_print_token if not no_stream else None),
         DownloadHttpFileTool(),
     ]
-    return Agent(model_name=model_name, tools=tools)
+    return Agent(
+        model_name=model_name, 
+        tools=tools,
+        compact_every_n_iterations=compact_every_n_iteration,
+        max_tokens_working_memory=max_tokens_working_memory,
+    )
 
 
-def create_full_agent(model_name: str, vision_model_name: str | None, no_stream: bool = False) -> Agent:
+def create_full_agent(
+    model_name: str, 
+    vision_model_name: str | None, 
+    no_stream: bool = False, 
+    compact_every_n_iteration: int | None = None,
+    max_tokens_working_memory: int | None = None
+) -> Agent:
     """Create an agent with the specified model and many tools.
 
     Args:
         model_name (str): Name of the model to use
         vision_model_name (str | None): Name of the vision model to use
         no_stream (bool, optional): If True, the agent will not stream results.
+        compact_every_n_iteration (int | None, optional): Frequency of memory compaction.
+        max_tokens_working_memory (int | None, optional): Maximum tokens for working memory.
 
     Returns:
         Agent: An agent with the specified model and tools
@@ -140,16 +171,26 @@ def create_full_agent(model_name: str, vision_model_name: str | None, no_stream:
     return Agent(
         model_name=model_name,
         tools=tools,
+        compact_every_n_iterations=compact_every_n_iteration,
+        max_tokens_working_memory=max_tokens_working_memory,
     )
 
 
-def create_orchestrator_agent(model_name: str, vision_model_name: str | None = None, no_stream: bool = False) -> Agent:
+def create_orchestrator_agent(
+    model_name: str, 
+    vision_model_name: str | None = None, 
+    no_stream: bool = False, 
+    compact_every_n_iteration: int | None = None,
+    max_tokens_working_memory: int | None = None
+) -> Agent:
     """Create an agent with the specified model and tools.
 
     Args:
         model_name (str): Name of the model to use
         vision_model_name (str | None): Name of the vision model to use
         no_stream (bool, optional): If True, the agent will not stream results.
+        compact_every_n_iteration (int | None, optional): Frequency of memory compaction.
+        max_tokens_working_memory (int | None, optional): Maximum tokens for working memory.
 
     Returns:
         Agent: An agent with the specified model and tools
@@ -175,4 +216,6 @@ def create_orchestrator_agent(model_name: str, vision_model_name: str | None = N
     return Agent(
         model_name=model_name,
         tools=tools,
+        compact_every_n_iterations=compact_every_n_iteration,
+        max_tokens_working_memory=max_tokens_working_memory,
     )

quantalogic/coding_agent.py

@@ -20,7 +20,14 @@ from quantalogic.utils import get_coding_environment
 from quantalogic.utils.get_quantalogic_rules_content import get_quantalogic_rules_file_content
 
 
-def create_coding_agent(model_name: str, vision_model_name: str | None = None, basic: bool = False,no_stream: bool = False) -> Agent:
+def create_coding_agent(
+    model_name: str, 
+    vision_model_name: str | None = None, 
+    basic: bool = False, 
+    no_stream: bool = False, 
+    compact_every_n_iteration: int | None = None,
+    max_tokens_working_memory: int | None = None
+) -> Agent:
     """Creates and configures a coding agent with a comprehensive set of tools.
 
     Args:
@@ -28,6 +35,8 @@ def create_coding_agent(model_name: str, vision_model_name: str | None = None, b
         vision_model_name (str | None): Name of the vision model to use for the agent's core capabilities
         basic (bool, optional): If True, the agent will be configured with a basic set of tools.
         no_stream (bool, optional): If True, the agent will not stream results.
+        compact_every_n_iteration (int | None, optional): Frequency of memory compaction.
+        max_tokens_working_memory (int | None, optional): Maximum tokens for working memory.
 
     Returns:
         Agent: A fully configured coding agent instance with:
@@ -91,4 +100,6 @@ def create_coding_agent(model_name: str, vision_model_name: str | None = None, b
         tools=tools,
         specific_expertise=specific_expertise,
         get_environment=get_coding_environment,
+        compact_every_n_iterations=compact_every_n_iteration,
+        max_tokens_working_memory=max_tokens_working_memory,
     )

quantalogic/generative_model.py

@@ -2,6 +2,7 @@
 
 import functools
 
+import litellm
 import openai
 from litellm import completion, exceptions, get_max_tokens, get_model_info, token_counter
 from loguru import logger
@@ -12,6 +13,11 @@ from quantalogic.event_emitter import EventEmitter  # Importing the EventEmitter
 MIN_RETRIES = 1
 
 
+
+litellm.suppress_debug_info = True # Very important to suppress prints don't remove
+
+
+
 # Define the Message class for conversation handling
 class Message(BaseModel):
     """Represents a message in a conversation with a specific role and content."""

quantalogic/main.py

@@ -11,7 +11,6 @@ import click
 from loguru import logger
 
 from quantalogic.console_print_events import console_print_events
-from quantalogic.console_print_token import console_print_token
 from quantalogic.utils.check_version import check_if_is_latest_version
 from quantalogic.version import get_version
 
@@ -23,7 +22,6 @@ from threading import Lock  # noqa: E402
 from rich.console import Console  # noqa: E402
 from rich.panel import Panel  # noqa: E402
 from rich.prompt import Confirm  # noqa: E402
-from rich.spinner import Spinner  # noqa: E402
 
 from quantalogic.agent import Agent  # noqa: E402
 
@@ -41,26 +39,28 @@ from quantalogic.search_agent import create_search_agent  # noqa: E402
 AGENT_MODES = ["code", "basic", "interpreter", "full", "code-basic", "search", "search-full"]
 
 
-def create_agent_for_mode(mode: str, model_name: str, vision_model_name: str | None, no_stream: bool = False) -> Agent:
+def create_agent_for_mode(mode: str, model_name: str, vision_model_name: str | None, no_stream: bool = False, compact_every_n_iteration: int | None = None, max_tokens_working_memory: int | None = None) -> Agent:
     """Create an agent based on the specified mode."""
     logger.debug(f"Creating agent for mode: {mode} with model: {model_name}")
     logger.debug(f"Using vision model: {vision_model_name}")
     logger.debug(f"Using no_stream: {no_stream}")
+    logger.debug(f"Using compact_every_n_iteration: {compact_every_n_iteration}")
+    logger.debug(f"Using max_tokens_working_memory: {max_tokens_working_memory}")
     if mode == "code":
         logger.debug("Creating code agent without basic mode")
-        return create_coding_agent(model_name, vision_model_name, basic=False, no_stream=no_stream)
+        return create_coding_agent(model_name, vision_model_name, basic=False, no_stream=no_stream, compact_every_n_iteration=compact_every_n_iteration, max_tokens_working_memory=max_tokens_working_memory)
     if mode == "code-basic":
-        return create_coding_agent(model_name, vision_model_name, basic=True, no_stream=no_stream)
+        return create_coding_agent(model_name, vision_model_name, basic=True, no_stream=no_stream, compact_every_n_iteration=compact_every_n_iteration, max_tokens_working_memory=max_tokens_working_memory)
     elif mode == "basic":
-        return create_orchestrator_agent(model_name, vision_model_name, no_stream=no_stream)
+        return create_orchestrator_agent(model_name, vision_model_name, no_stream=no_stream, compact_every_n_iteration=compact_every_n_iteration, max_tokens_working_memory=max_tokens_working_memory)
     elif mode == "full":
-        return create_full_agent(model_name, vision_model_name, no_stream=no_stream)
+        return create_full_agent(model_name, vision_model_name, no_stream=no_stream, compact_every_n_iteration=compact_every_n_iteration, max_tokens_working_memory=max_tokens_working_memory)
     elif mode == "interpreter":
-        return create_interpreter_agent(model_name, vision_model_name, no_stream=no_stream)
+        return create_interpreter_agent(model_name, vision_model_name, no_stream=no_stream, compact_every_n_iteration=compact_every_n_iteration, max_tokens_working_memory=max_tokens_working_memory)
     elif mode == "search":
-        return create_search_agent(model_name, no_stream=no_stream)
+        return create_search_agent(model_name, no_stream=no_stream, compact_every_n_iteration=compact_every_n_iteration, max_tokens_working_memory=max_tokens_working_memory)
     if mode == "search-full":
-        return create_search_agent(model_name, mode_full=True, no_stream=no_stream)
+        return create_search_agent(model_name, mode_full=True, no_stream=no_stream, compact_every_n_iteration=compact_every_n_iteration, max_tokens_working_memory=max_tokens_working_memory)
     else:
         raise ValueError(f"Unknown agent mode: {mode}")
 
@@ -154,7 +154,12 @@ def stop_spinner(console: Console) -> None:
 
 
 def display_welcome_message(
-    console: Console, model_name: str, vision_model_name: str | None, max_iterations: int = 50
+    console: Console, 
+    model_name: str, 
+    vision_model_name: str | None, 
+    max_iterations: int = 50, 
+    compact_every_n_iteration: int | None = None,
+    max_tokens_working_memory: int | None = None
 ) -> None:
     """Display the welcome message and instructions."""
     version = get_version()
@@ -169,7 +174,9 @@ def display_welcome_message(
             "\n"
             f"- Model: {model_name}\n"
             f"- Vision Model: {vision_model_name}\n"
-            f"- Max Iterations: {max_iterations}\n\n"
+            f"- Max Iterations: {max_iterations}\n"
+            f"- Memory Compact Frequency: {compact_every_n_iteration or 'Default (Max Iterations)'}\n"
+            f"- Max Working Memory Tokens: {max_tokens_working_memory or 'Default'}\n\n"
             "[bold magenta]💡 Pro Tips:[/bold magenta]\n\n"
             "- Be as specific as possible in your task description to get the best results!\n"
             "- Use clear and concise language when describing your task\n"
@@ -182,6 +189,12 @@ def display_welcome_message(
 
 
 @click.group(invoke_without_command=True)
+@click.option(
+    "--compact-every-n-iteration",
+    type=int,
+    default=None,
+    help="Set the frequency of memory compaction for the agent (default: max_iterations)."
+)
 @click.option("--version", is_flag=True, help="Show version information.")
 @click.option(
     "--model-name",
@@ -207,6 +220,12 @@ def display_welcome_message(
     default=30,
     help="Maximum number of iterations for task solving (default: 30).",
 )
+@click.option(
+    "--max-tokens-working-memory",
+    type=int,
+    default=None,
+    help="Set the maximum number of tokens allowed in the working memory."
+)
 @click.pass_context
 def cli(
     ctx: click.Context,
@@ -217,6 +236,8 @@ def cli(
     log: str,
     vision_model_name: str | None,
     max_iterations: int,
+    compact_every_n_iteration: int | None,
+    max_tokens_working_memory: int | None,
 ) -> None:
     """QuantaLogic AI Assistant - A powerful AI tool for various tasks."""
     if version:
@@ -232,6 +253,8 @@ def cli(
             log=log,
             vision_model_name=vision_model_name,
             max_iterations=max_iterations,
+            compact_every_n_iteration=compact_every_n_iteration,
+            max_tokens_working_memory=max_tokens_working_memory,
         )
 
 
@@ -261,6 +284,18 @@ def cli(
     default=30,
     help="Maximum number of iterations for task solving (default: 30).",
 )
+@click.option(
+    "--compact-every-n-iteration",
+    type=int,
+    default=None,
+    help="Set the frequency of memory compaction for the agent (default: max_iterations)."
+)
+@click.option(
+    "--max-tokens-working-memory",
+    type=int,
+    default=None,
+    help="Set the maximum number of tokens allowed in the working memory."
+)
 @click.option(
     "--no-stream",
     is_flag=True,
@@ -276,6 +311,8 @@ def task(
     vision_model_name: str | None,
     task: Optional[str],
     max_iterations: int,
+    compact_every_n_iteration: int | None,
+    max_tokens_working_memory: int | None,
     no_stream: bool,
 ) -> None:
     """Execute a task with the QuantaLogic AI Assistant."""
@@ -290,7 +327,14 @@ def task(
                 check_new_version()
                 task_content = task
             else:
-                display_welcome_message(console, model_name, vision_model_name, max_iterations=max_iterations)
+                display_welcome_message(
+                    console, 
+                    model_name, 
+                    vision_model_name, 
+                    max_iterations=max_iterations, 
+                    compact_every_n_iteration=compact_every_n_iteration,
+                    max_tokens_working_memory=max_tokens_working_memory
+                )
                 check_new_version()
                 logger.debug("Waiting for user input...")
                 task_content = get_multiline_input(console).strip()
@@ -322,7 +366,7 @@ def task(
         logger.debug(
             f"Creating agent for mode: {mode} with model: {model_name}, vision model: {vision_model_name}, no_stream: {no_stream}"
         )
-        agent = create_agent_for_mode(mode, model_name, vision_model_name=vision_model_name, no_stream=no_stream)
+        agent = create_agent_for_mode(mode, model_name, vision_model_name=vision_model_name, no_stream=no_stream, compact_every_n_iteration=compact_every_n_iteration, max_tokens_working_memory=max_tokens_working_memory)
         logger.debug(
             f"Created agent for mode: {mode} with model: {model_name}, vision model: {vision_model_name}, no_stream: {no_stream}"
         )

quantalogic/prompts.py

@@ -17,7 +17,7 @@ Every response must contain exactly two XML blocks:
 1. Analysis Block:
 ```xml
 <thinking>
-  <!-- You must follow this precise format, be very concise and very precise -->
+  <!-- Must follow this precise format, concise, dense, use abreviations, emojis, unicode characters to make it denser -->
  <task_analysis_if_no_history> 
    Only if no conversation history:
     * Rewrite the <task> and its context with your own words in detailed, clear, and specific manner.
@@ -51,7 +51,7 @@ Every response must contain exactly two XML blocks:
   </last_observation>
   <progess_analysis>
      <!-- if there is a conversation history -->
-    * Detail each step failed and completed so far.
+    * Detail each step failed and completed so far, be concise.
     * Identify and evaluate any blockers or challenges to the progress of global task.
     * Identify repetitions: if repeated steps, take a step back and rethink your approach.
     * Provide potential solutions, and if needed, suggest reevaluating the approach and the plan.

quantalogic/search_agent.py

@@ -12,12 +12,21 @@ from quantalogic.tools import (
 )
 
 
-def create_search_agent(model_name: str, mode_full: bool = False) -> Agent:
+def create_search_agent(
+    model_name: str, 
+    mode_full: bool = False, 
+    no_stream: bool = False, 
+    compact_every_n_iteration: int | None = None,
+    max_tokens_working_memory: int | None = None
+) -> Agent:
     """Creates and configures a search agent with web, knowledge, and privacy-focused search tools.
 
     Args:
         model_name (str): Name of the language model to use for the agent's core capabilities
         mode_full (bool, optional): If True, the agent will be configured with a full set of tools.
+        no_stream (bool, optional): If True, the agent will not stream results.
+        compact_every_n_iteration (int | None, optional): Frequency of memory compaction.
+        max_tokens_working_memory (int | None, optional): Maximum tokens for working memory.
 
     Returns:
         Agent: A fully configured search agent instance with:
@@ -57,4 +66,6 @@ def create_search_agent(model_name: str, mode_full: bool = False) -> Agent:
         model_name=model_name,
         tools=tools,
         specific_expertise=specific_expertise,
+        compact_every_n_iterations=compact_every_n_iteration,
+        max_tokens_working_memory=max_tokens_working_memory,
     )

{quantalogic-0.2.17.dist-info → quantalogic-0.2.18.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: quantalogic
-Version: 0.2.17
+Version: 0.2.18
 Summary: QuantaLogic ReAct Agents
 Author: Raphaël MANSUY
 Author-email: raphael.mansuy@gmail.com
@@ -81,6 +81,8 @@ We created [QuantaLogic](https://www.quantalogic.app) because we saw a significa
 
 ## 📋 Table of Contents
 
+- [Release Notes](#release-notes)
+
 - [Installation](#-installation)
 - [Quick Start](#-quickstart)
 - [Key Components](#-key-components)
@@ -93,6 +95,12 @@ We created [QuantaLogic](https://www.quantalogic.app) because we saw a significa
 - [License](#-license)
 - [Documentation Development](#-documentation-development)
 
+## Release Notes
+
+See our [Release Notes](RELEASE_NOTES.MD) for detailed version history and changes.
+
+[TODO List](TODO.md)
+
 ## 📦 Installation
 
 ### Prerequisites
@@ -134,20 +142,18 @@ The QuantaLogic CLI provides powerful command-line capabilities:
 
 ```bash
 Usage: quantalogic [OPTIONS] COMMAND [ARGS]...
-
   QuantaLogic AI Assistant - A powerful AI tool for various tasks.
 
 Options:
   --version                       Show version information.
-  --model-name TEXT               Specify the text model to use (litellm format,
-                                  e.g. "openrouter/deepseek/deepseek-chat").
-  --vision-model-name TEXT        Specify the vision model to use (litellm format,
-                                  e.g. "openrouter/openai/gpt-4o-mini").
-  --log [info|debug|warning]      Set logging level (info/debug/warning).
+  --model-name TEXT               Specify the model (litellm format, e.g., "openrouter/deepseek/deepseek-chat").
+  --log [info|debug|warning]      Set logging level.
   --verbose                       Enable verbose output.
+  --mode [code|basic|interpreter|full|code-basic|search|search-full]  Agent mode.
+  --vision-model-name TEXT        Specify the vision model (litellm format, e.g., "openrouter/A/gpt-4o-mini").
   --max-iterations INTEGER        Maximum iterations for task solving (default: 30).
-  --mode [code|basic|interpreter|full|code-basic|search|search-full]
-                                  Agent mode (code/search/full).
+  --max-tokens-working-memory INTEGER Maximum tokens to keep in working memory (default: 4000).
+  --compact-every-n-iteration INTEGER Compact memory every N iterations (default: 5).
   --help                          Show this message and exit.
 
 Commands:
@@ -157,6 +163,38 @@ Commands:
 ### Commands
 task    Execute a task with the QuantaLogic AI Assistant
 
+**Usage:** `quantalogic task [OPTIONS] [TASK]`  
+**Description:** Execute a task with the QuantaLogic AI Assistant.  
+**Options:**  
+- `--file PATH`: Path to task file.  
+- `--model-name TEXT`: Specify the model (litellm format, e.g., `openrouter/deepseek/deepseek-chat`).  
+- `--verbose`: Enable verbose output.  
+- `--mode [code|basic|interpreter|full|code-basic|search|search-full]`: Agent mode.  
+- `--log [info|debug|warning]`: Set logging level.  
+- `--vision-model-name TEXT`: Specify the vision model (litellm format).  
+- `--max-iterations INTEGER`: Maximum iterations for task solving (default: 30).  
+- `--max-tokens-working-memory INTEGER`: Maximum tokens to keep in working memory (default: 4000).  
+- `--compact-every-n-iteration INTEGER`: Compact memory every N iterations (default: 5).  
+- `--no-stream`: Disable streaming output (default: enabled).  
+- `--help`: Show this message and exit.
+
+**Detailed Parameter Descriptions:**
+
+- **--model-name**: Specifies the LLM model to use (e.g., "openrouter/deepseek/deepseek-chat")
+- **--mode**: Selects agent capabilities:
+  - *code*: Coding-focused with basic capabilities
+  - *basic*: General-purpose without coding tools
+  - *interpreter*: Interactive code execution
+  - *full*: All capabilities enabled
+  - *code-basic*: Coding with basic reasoning
+  - *search*: Web search integration
+- **--log**: Controls logging verbosity (info, debug, warning)
+- **--vision-model-name**: Specifies vision model for image processing
+- **--max-iterations**: Limits task-solving attempts (default: 30)
+- **--max-tokens-working-memory**: Controls memory usage (default: None)
+- **--compact-every-n-iteration**: Memory optimization frequency (default: None)
+- **--no-stream**: Disables real-time output streaming
+
 
 
 ### Detailed Usage

{quantalogic-0.2.17.dist-info → quantalogic-0.2.18.dist-info}/RECORD

@@ -1,18 +1,18 @@
 quantalogic/__init__.py,sha256=kX0c_xmD9OslWnAE92YHMGuD7xZcTo8ZOF_5R64HKps,784
-quantalogic/agent.py,sha256=w4ljI0ccOagTioJLFWQH05M7P304qe-NB6Cca_JIjgU,27900
-quantalogic/agent_config.py,sha256=zS7uPPPJ7sxKy4iBY5lkgDdDZR7H6mEGy4Q605-4oXs,5516
-quantalogic/coding_agent.py,sha256=NDXZyGBwxnjy3E5MmZ43axA1IIq7uFH14DpbNRVLWzM,3899
+quantalogic/agent.py,sha256=KiONndkTQ_xTjlmwA2NcFwt4VGlBxSt40yHrL8Xrlms,29997
+quantalogic/agent_config.py,sha256=H-YkaNkVglBlnN00Htrs3bPPoD2hirBfGBpoSbKgdMU,7232
+quantalogic/coding_agent.py,sha256=viKePGtDbD3tzbc6RIPslINeRqOkLYwDWXDsw4TF18Y,4328
 quantalogic/console_print_events.py,sha256=KB-DGi52As8M96eUs1N_vgNqKIFtqv_H8NTOd3TLTgQ,2163
 quantalogic/console_print_token.py,sha256=qSU-3kmoZk4T5-1ybrEBi8tIXDPcz7eyWKhGh3E8uIg,395
 quantalogic/docs_cli.py,sha256=3giVbUpespB9ZdTSJ955A3BhcOaBl5Lwsn1AVy9XAeY,1663
 quantalogic/event_emitter.py,sha256=jqot2g4JRXc88K6PW837Oqxbf7shZfO-xdPaUWmzupk,7901
-quantalogic/generative_model.py,sha256=DcLFYNS9H588dPfIL5s4pajG9EVpdPN0Mp36waQIFRk,12294
+quantalogic/generative_model.py,sha256=s94heYFPSA2IO4E3ay3CF1k9hXKNAzX1KVmxHlQgF_8,12397
 quantalogic/interactive_text_editor.py,sha256=kYeTA2qej5kxtPvAUHy_Dr2MhrGQAyenLFpW9mU9Rmw,6855
-quantalogic/main.py,sha256=rHctMiV9-XRf6O-PIisjOdaF2y1SQNclUJQ53KPdh9A,14138
+quantalogic/main.py,sha256=nyqwQwrzmiurCwR5gkkx16BNW9XN69fTHvqOZ_qK9CU,16655
 quantalogic/memory.py,sha256=zbtRuM05jaS2lJll-92dt5JfYVLERnF_m_9xqp2x-k0,6304
 quantalogic/model_names.py,sha256=UZlz25zG9B2dpfwdw_e1Gw5qFsKQ7iME9FJh9Ts4u6s,938
-quantalogic/prompts.py,sha256=jRAm9YYvwx8KAx42Ak2H7y2p_1jIVuL1e3Wsd2Y81Cc,3548
-quantalogic/search_agent.py,sha256=Jy-Zw77giZOlON_bDO0u3ppfRxv_O0EGcGmN4QuuLmQ,1944
+quantalogic/prompts.py,sha256=CW4CRgW1hTpXeWdeJNbPaRPUeUm-xKuGHJrT8mOtvkw,3602
+quantalogic/search_agent.py,sha256=EA_FAPP0dVuUbJ_lAGKfYq1FIJ6oLYzGMgKLMvBL4ZQ,2472
 quantalogic/server/__init__.py,sha256=8sz_PYAUCrkM6JM5EAUeIzNM4NPW6j6UT72JVkc21WQ,91
 quantalogic/server/agent_server.py,sha256=JNginXlkgBfJLqgKffoTBSvBWI8ZO1uXUgTOqLINk_g,22517
 quantalogic/server/models.py,sha256=nVUGWElOsUw8QnRCGJylk25wCew_5gohe6nldYighUA,1322
@@ -70,8 +70,8 @@ quantalogic/utils/read_http_text_content.py,sha256=n3IayT5KcqctIVVF2gOQQAMf3Ow6e
 quantalogic/version.py,sha256=ea_cRutaQk5_lwlLbUUvPFuOT7Of7-gAsDl7wdveS-g,107
 quantalogic/xml_parser.py,sha256=uMLQNHTRCg116FwcjRoquZmSwVtE4LEH-6V2E3RD-dA,11466
 quantalogic/xml_tool_parser.py,sha256=lsVzClZBrZan7wjCuCKnGHWzksXI3VMy_vWthxu2_bo,3738
-quantalogic-0.2.17.dist-info/LICENSE,sha256=xx0jnfkXJvxRnG63LTGOxlggYnIysveWIZ6H3PNdCrQ,11357
-quantalogic-0.2.17.dist-info/METADATA,sha256=0RCGjrnGzipu7eGuYGEgmd1jDFWetzPCjzHHBMHDS_o,38574
-quantalogic-0.2.17.dist-info/WHEEL,sha256=Nq82e9rUAnEjt98J6MlVmMCZb-t9cYE2Ir1kpBmnWfs,88
-quantalogic-0.2.17.dist-info/entry_points.txt,sha256=h74O_Q3qBRCrDR99qvwB4BpBGzASPUIjCfxHq6Qnups,183
-quantalogic-0.2.17.dist-info/RECORD,,
+quantalogic-0.2.18.dist-info/LICENSE,sha256=xx0jnfkXJvxRnG63LTGOxlggYnIysveWIZ6H3PNdCrQ,11357
+quantalogic-0.2.18.dist-info/METADATA,sha256=W9clERsCPJHiuK25qx_dhiS3eitzGjzNHl8eA99kN2s,40514
+quantalogic-0.2.18.dist-info/WHEEL,sha256=Nq82e9rUAnEjt98J6MlVmMCZb-t9cYE2Ir1kpBmnWfs,88
+quantalogic-0.2.18.dist-info/entry_points.txt,sha256=h74O_Q3qBRCrDR99qvwB4BpBGzASPUIjCfxHq6Qnups,183
+quantalogic-0.2.18.dist-info/RECORD,,