open-research-lite 0.1.0__tar.gz

open_research_lite-0.1.0/LICENSE

@@ -0,0 +1,21 @@
+MIT License
+
+Copyright (c) 2025 LangChain
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.

open_research_lite-0.1.0/PKG-INFO

@@ -0,0 +1,156 @@
+Metadata-Version: 2.4
+Name: open-research-lite
+Version: 0.1.0
+Summary: Token-Efficient Ingestion Middleware & Concept-Diff Engine for Deep Research AI Agents.
+Author: DeepResearch-Lite Team
+License: MIT
+Requires-Python: >=3.10
+Description-Content-Type: text/markdown
+License-File: LICENSE
+Requires-Dist: langgraph>=0.5.4
+Requires-Dist: langchain-community>=0.3.9
+Requires-Dist: langchain-openai>=0.3.28
+Requires-Dist: langchain-anthropic>=0.3.15
+Requires-Dist: langchain-mcp-adapters>=0.1.6
+Requires-Dist: langchain-tavily
+Requires-Dist: openai>=1.99.2
+Requires-Dist: tavily-python>=0.5.0
+Requires-Dist: pydantic>=2.0.0
+Requires-Dist: requests>=2.32.3
+Requires-Dist: beautifulsoup4==4.14.3
+Requires-Dist: python-dotenv>=1.0.1
+Requires-Dist: pytest
+Requires-Dist: langchain-google-genai>=2.1.5
+Provides-Extra: dev
+Requires-Dist: mypy>=1.11.1; extra == "dev"
+Requires-Dist: ruff>=0.6.1; extra == "dev"
+Dynamic: license-file
+
+# DeepResearch-Lite ⚡
+> **Token-Efficient Ingestion Middleware & Concept-Diff Engine for Deep Research AI Agents**
+
+[![License: MIT](https://img.shields.io/badge/License-MIT-yellow.svg)](https://opensource.org/licenses/MIT)
+[![Python 3.10+](https://img.shields.io/badge/python-3.10+-blue.svg)](https://www.python.org/downloads/)
+[![Token Savings](https://img.shields.io/badge/Token_Savings-75%25--85%25-green.svg)](#benchmark--scorecard)
+
+`DeepResearch-Lite` is an intelligent, high-performance gatekeeper layer designed for AI Research Agents (such as `langchain-ai/open_deep_research`, `smolagents`, `AutoGPT`, `CrewAI`, or custom LangGraph workflows). 
+
+It solves the **Extreme Token Bloat & Repetitive Fluff** flaw in existing deep research systems by shifting the paradigm from *"Read raw 5,000-word scraped web pages"* to **"Compile knowledge incrementally as it arrives."**
+
+---
+
+## 🚀 Key Features
+
+* **Dual-Layer Extraction Architecture**:
+  - **Layer 1 (LLM Mode)**: Sub-second structured JSON triplet extraction `(Subject ──► Predicate ──► Object)` via **Gemini 2.5 Flash** or **OpenAI Mini** when an API key is available.
+  - **Layer 2 (Professional Local NLP Mode)**: 0-ms offline local NLP parser (regex metric extraction, grammar triplet matching, entity resolution) when running without API keys.
+* **Session Knowledge Graph**: Maintains an in-memory graph state of all entities, metrics, and claims learned during a research session.
+* **Concept-Diff Engine**: Classifies scraped facts into 3 categories:
+  - `DISCARD`: Repetitive background fluff deleted immediately ($0 LLM tokens spent).
+  - `DIFF_ADD`: Novel assertions added to graph & passed in Diff Payload.
+  - `DIFF_CONFLICT`: Contradictory statements (e.g. $110 vs $140/kWh) explicitly flagged.
+* **80%+ Token & Cost Reduction**: Delivers a ~150-word **Diff Payload** to the main reasoning model instead of 5,000 words of background history.
+
+---
+
+## ⚡ Quickstart
+
+### 1. Installation
+
+```bash
+pip install -e .
+```
+
+### 2. Basic Usage (Python API)
+
+```python
+import asyncio
+from open_deep_research.concept_diff import ConceptDiffEngine
+
+async def main():
+    # Automatically uses GEMINI_API_KEY if present, otherwise uses Professional Local NLP
+    engine = ConceptDiffEngine()
+
+    raw_scraped_text = """
+    Electric vehicles have become popular over the last decade... 
+    Lithium-ion batteries were invented by John Goodenough...
+    In 2026, researchers demonstrated a solid-state cell achieving 500 Wh/kg energy density.
+    Target production cell cost is $110/kWh.
+    """
+
+    # Process raw scrape into a condensed Diff Payload
+    diff_payload = await engine.process_observation(
+        raw_text=raw_scraped_text,
+        source_url="https://tech-news.com/ev-batteries",
+        source_title="EV Battery 2026 Report"
+    )
+
+    print(diff_payload)
+
+if __name__ == "__main__":
+    asyncio.run(main())
+```
+
+---
+
+## 📊 Benchmark & Scorecard
+
+Running `DeepResearch-Lite` on real web searches (`2026 solid state battery Wh/kg breakthroughs`):
+
+```
+===========================================================================
+🏆 DEEPRESEARCH-LITE HACKATHON SCORECARD
+===========================================================================
+METRIC                         | BASELINE AGENT     | DEEPRESEARCH-LITE 
+---------------------------------------------------------------------------
+Live Web Words Fetched         | 9,159 words        | 1,083 words       
+Input Tokens Passed to LLM     | 11,906 tokens      | 1,407 tokens      
+Estimated Cost per Search Run  | $0.0298            | $0.0035           
+Token Reduction                | 0% (Full Bloat)    | 88.2% SAVED ⚡     
+Fact-to-Fluff Signal Ratio     | ~15% High Signal   | ~95% High Signal ⚡
+===========================================================================
+```
+
+---
+
+## 🛠️ Framework Integration
+
+### LangGraph / open_deep_research Integration
+`DeepResearch-Lite` integrates seamlessly into `open_deep_research/deep_researcher.py` at the `researcher_tools()` node:
+
+```python
+from open_deep_research.concept_diff import ConceptDiffEngine
+
+diff_engine = ConceptDiffEngine()
+
+# In researcher_tools():
+processed_observations = []
+for obs, tool_call in zip(observations, tool_calls):
+    if len(obs) > 100:
+        diff_payload = await diff_engine.process_observation(obs, source_title=tool_call['name'])
+        processed_observations.append(diff_payload)
+```
+
+---
+
+## 🧪 Testing & Demonstration Scripts
+
+- **Run Unit Tests**:
+  ```bash
+  python -m pytest tests/test_concept_diff.py
+  ```
+
+- **Run Side-by-Side Hackathon Benchmark**:
+  ```bash
+  python demo_comparison.py
+  ```
+
+- **Run Live Internet Search Comparison**:
+  ```bash
+  python run_live_test.py "your custom research prompt"
+  ```
+
+---
+
+## 📄 License
+MIT License. Free for open-source and commercial use.

open_research_lite-0.1.0/README.md

@@ -0,0 +1,128 @@
+# DeepResearch-Lite ⚡
+> **Token-Efficient Ingestion Middleware & Concept-Diff Engine for Deep Research AI Agents**
+
+[![License: MIT](https://img.shields.io/badge/License-MIT-yellow.svg)](https://opensource.org/licenses/MIT)
+[![Python 3.10+](https://img.shields.io/badge/python-3.10+-blue.svg)](https://www.python.org/downloads/)
+[![Token Savings](https://img.shields.io/badge/Token_Savings-75%25--85%25-green.svg)](#benchmark--scorecard)
+
+`DeepResearch-Lite` is an intelligent, high-performance gatekeeper layer designed for AI Research Agents (such as `langchain-ai/open_deep_research`, `smolagents`, `AutoGPT`, `CrewAI`, or custom LangGraph workflows). 
+
+It solves the **Extreme Token Bloat & Repetitive Fluff** flaw in existing deep research systems by shifting the paradigm from *"Read raw 5,000-word scraped web pages"* to **"Compile knowledge incrementally as it arrives."**
+
+---
+
+## 🚀 Key Features
+
+* **Dual-Layer Extraction Architecture**:
+  - **Layer 1 (LLM Mode)**: Sub-second structured JSON triplet extraction `(Subject ──► Predicate ──► Object)` via **Gemini 2.5 Flash** or **OpenAI Mini** when an API key is available.
+  - **Layer 2 (Professional Local NLP Mode)**: 0-ms offline local NLP parser (regex metric extraction, grammar triplet matching, entity resolution) when running without API keys.
+* **Session Knowledge Graph**: Maintains an in-memory graph state of all entities, metrics, and claims learned during a research session.
+* **Concept-Diff Engine**: Classifies scraped facts into 3 categories:
+  - `DISCARD`: Repetitive background fluff deleted immediately ($0 LLM tokens spent).
+  - `DIFF_ADD`: Novel assertions added to graph & passed in Diff Payload.
+  - `DIFF_CONFLICT`: Contradictory statements (e.g. $110 vs $140/kWh) explicitly flagged.
+* **80%+ Token & Cost Reduction**: Delivers a ~150-word **Diff Payload** to the main reasoning model instead of 5,000 words of background history.
+
+---
+
+## ⚡ Quickstart
+
+### 1. Installation
+
+```bash
+pip install -e .
+```
+
+### 2. Basic Usage (Python API)
+
+```python
+import asyncio
+from open_deep_research.concept_diff import ConceptDiffEngine
+
+async def main():
+    # Automatically uses GEMINI_API_KEY if present, otherwise uses Professional Local NLP
+    engine = ConceptDiffEngine()
+
+    raw_scraped_text = """
+    Electric vehicles have become popular over the last decade... 
+    Lithium-ion batteries were invented by John Goodenough...
+    In 2026, researchers demonstrated a solid-state cell achieving 500 Wh/kg energy density.
+    Target production cell cost is $110/kWh.
+    """
+
+    # Process raw scrape into a condensed Diff Payload
+    diff_payload = await engine.process_observation(
+        raw_text=raw_scraped_text,
+        source_url="https://tech-news.com/ev-batteries",
+        source_title="EV Battery 2026 Report"
+    )
+
+    print(diff_payload)
+
+if __name__ == "__main__":
+    asyncio.run(main())
+```
+
+---
+
+## 📊 Benchmark & Scorecard
+
+Running `DeepResearch-Lite` on real web searches (`2026 solid state battery Wh/kg breakthroughs`):
+
+```
+===========================================================================
+🏆 DEEPRESEARCH-LITE HACKATHON SCORECARD
+===========================================================================
+METRIC                         | BASELINE AGENT     | DEEPRESEARCH-LITE 
+---------------------------------------------------------------------------
+Live Web Words Fetched         | 9,159 words        | 1,083 words       
+Input Tokens Passed to LLM     | 11,906 tokens      | 1,407 tokens      
+Estimated Cost per Search Run  | $0.0298            | $0.0035           
+Token Reduction                | 0% (Full Bloat)    | 88.2% SAVED ⚡     
+Fact-to-Fluff Signal Ratio     | ~15% High Signal   | ~95% High Signal ⚡
+===========================================================================
+```
+
+---
+
+## 🛠️ Framework Integration
+
+### LangGraph / open_deep_research Integration
+`DeepResearch-Lite` integrates seamlessly into `open_deep_research/deep_researcher.py` at the `researcher_tools()` node:
+
+```python
+from open_deep_research.concept_diff import ConceptDiffEngine
+
+diff_engine = ConceptDiffEngine()
+
+# In researcher_tools():
+processed_observations = []
+for obs, tool_call in zip(observations, tool_calls):
+    if len(obs) > 100:
+        diff_payload = await diff_engine.process_observation(obs, source_title=tool_call['name'])
+        processed_observations.append(diff_payload)
+```
+
+---
+
+## 🧪 Testing & Demonstration Scripts
+
+- **Run Unit Tests**:
+  ```bash
+  python -m pytest tests/test_concept_diff.py
+  ```
+
+- **Run Side-by-Side Hackathon Benchmark**:
+  ```bash
+  python demo_comparison.py
+  ```
+
+- **Run Live Internet Search Comparison**:
+  ```bash
+  python run_live_test.py "your custom research prompt"
+  ```
+
+---
+
+## 📄 License
+MIT License. Free for open-source and commercial use.

open_research_lite-0.1.0/open_research_lite.egg-info/PKG-INFO

@@ -0,0 +1,156 @@
+Metadata-Version: 2.4
+Name: open-research-lite
+Version: 0.1.0
+Summary: Token-Efficient Ingestion Middleware & Concept-Diff Engine for Deep Research AI Agents.
+Author: DeepResearch-Lite Team
+License: MIT
+Requires-Python: >=3.10
+Description-Content-Type: text/markdown
+License-File: LICENSE
+Requires-Dist: langgraph>=0.5.4
+Requires-Dist: langchain-community>=0.3.9
+Requires-Dist: langchain-openai>=0.3.28
+Requires-Dist: langchain-anthropic>=0.3.15
+Requires-Dist: langchain-mcp-adapters>=0.1.6
+Requires-Dist: langchain-tavily
+Requires-Dist: openai>=1.99.2
+Requires-Dist: tavily-python>=0.5.0
+Requires-Dist: pydantic>=2.0.0
+Requires-Dist: requests>=2.32.3
+Requires-Dist: beautifulsoup4==4.14.3
+Requires-Dist: python-dotenv>=1.0.1
+Requires-Dist: pytest
+Requires-Dist: langchain-google-genai>=2.1.5
+Provides-Extra: dev
+Requires-Dist: mypy>=1.11.1; extra == "dev"
+Requires-Dist: ruff>=0.6.1; extra == "dev"
+Dynamic: license-file
+
+# DeepResearch-Lite ⚡
+> **Token-Efficient Ingestion Middleware & Concept-Diff Engine for Deep Research AI Agents**
+
+[![License: MIT](https://img.shields.io/badge/License-MIT-yellow.svg)](https://opensource.org/licenses/MIT)
+[![Python 3.10+](https://img.shields.io/badge/python-3.10+-blue.svg)](https://www.python.org/downloads/)
+[![Token Savings](https://img.shields.io/badge/Token_Savings-75%25--85%25-green.svg)](#benchmark--scorecard)
+
+`DeepResearch-Lite` is an intelligent, high-performance gatekeeper layer designed for AI Research Agents (such as `langchain-ai/open_deep_research`, `smolagents`, `AutoGPT`, `CrewAI`, or custom LangGraph workflows). 
+
+It solves the **Extreme Token Bloat & Repetitive Fluff** flaw in existing deep research systems by shifting the paradigm from *"Read raw 5,000-word scraped web pages"* to **"Compile knowledge incrementally as it arrives."**
+
+---
+
+## 🚀 Key Features
+
+* **Dual-Layer Extraction Architecture**:
+  - **Layer 1 (LLM Mode)**: Sub-second structured JSON triplet extraction `(Subject ──► Predicate ──► Object)` via **Gemini 2.5 Flash** or **OpenAI Mini** when an API key is available.
+  - **Layer 2 (Professional Local NLP Mode)**: 0-ms offline local NLP parser (regex metric extraction, grammar triplet matching, entity resolution) when running without API keys.
+* **Session Knowledge Graph**: Maintains an in-memory graph state of all entities, metrics, and claims learned during a research session.
+* **Concept-Diff Engine**: Classifies scraped facts into 3 categories:
+  - `DISCARD`: Repetitive background fluff deleted immediately ($0 LLM tokens spent).
+  - `DIFF_ADD`: Novel assertions added to graph & passed in Diff Payload.
+  - `DIFF_CONFLICT`: Contradictory statements (e.g. $110 vs $140/kWh) explicitly flagged.
+* **80%+ Token & Cost Reduction**: Delivers a ~150-word **Diff Payload** to the main reasoning model instead of 5,000 words of background history.
+
+---
+
+## ⚡ Quickstart
+
+### 1. Installation
+
+```bash
+pip install -e .
+```
+
+### 2. Basic Usage (Python API)
+
+```python
+import asyncio
+from open_deep_research.concept_diff import ConceptDiffEngine
+
+async def main():
+    # Automatically uses GEMINI_API_KEY if present, otherwise uses Professional Local NLP
+    engine = ConceptDiffEngine()
+
+    raw_scraped_text = """
+    Electric vehicles have become popular over the last decade... 
+    Lithium-ion batteries were invented by John Goodenough...
+    In 2026, researchers demonstrated a solid-state cell achieving 500 Wh/kg energy density.
+    Target production cell cost is $110/kWh.
+    """
+
+    # Process raw scrape into a condensed Diff Payload
+    diff_payload = await engine.process_observation(
+        raw_text=raw_scraped_text,
+        source_url="https://tech-news.com/ev-batteries",
+        source_title="EV Battery 2026 Report"
+    )
+
+    print(diff_payload)
+
+if __name__ == "__main__":
+    asyncio.run(main())
+```
+
+---
+
+## 📊 Benchmark & Scorecard
+
+Running `DeepResearch-Lite` on real web searches (`2026 solid state battery Wh/kg breakthroughs`):
+
+```
+===========================================================================
+🏆 DEEPRESEARCH-LITE HACKATHON SCORECARD
+===========================================================================
+METRIC                         | BASELINE AGENT     | DEEPRESEARCH-LITE 
+---------------------------------------------------------------------------
+Live Web Words Fetched         | 9,159 words        | 1,083 words       
+Input Tokens Passed to LLM     | 11,906 tokens      | 1,407 tokens      
+Estimated Cost per Search Run  | $0.0298            | $0.0035           
+Token Reduction                | 0% (Full Bloat)    | 88.2% SAVED ⚡     
+Fact-to-Fluff Signal Ratio     | ~15% High Signal   | ~95% High Signal ⚡
+===========================================================================
+```
+
+---
+
+## 🛠️ Framework Integration
+
+### LangGraph / open_deep_research Integration
+`DeepResearch-Lite` integrates seamlessly into `open_deep_research/deep_researcher.py` at the `researcher_tools()` node:
+
+```python
+from open_deep_research.concept_diff import ConceptDiffEngine
+
+diff_engine = ConceptDiffEngine()
+
+# In researcher_tools():
+processed_observations = []
+for obs, tool_call in zip(observations, tool_calls):
+    if len(obs) > 100:
+        diff_payload = await diff_engine.process_observation(obs, source_title=tool_call['name'])
+        processed_observations.append(diff_payload)
+```
+
+---
+
+## 🧪 Testing & Demonstration Scripts
+
+- **Run Unit Tests**:
+  ```bash
+  python -m pytest tests/test_concept_diff.py
+  ```
+
+- **Run Side-by-Side Hackathon Benchmark**:
+  ```bash
+  python demo_comparison.py
+  ```
+
+- **Run Live Internet Search Comparison**:
+  ```bash
+  python run_live_test.py "your custom research prompt"
+  ```
+
+---
+
+## 📄 License
+MIT License. Free for open-source and commercial use.

open_research_lite-0.1.0/open_research_lite.egg-info/SOURCES.txt

@@ -0,0 +1,28 @@
+LICENSE
+README.md
+pyproject.toml
+open_research_lite.egg-info/PKG-INFO
+open_research_lite.egg-info/SOURCES.txt
+open_research_lite.egg-info/dependency_links.txt
+open_research_lite.egg-info/requires.txt
+open_research_lite.egg-info/top_level.txt
+src/legacy/__init__.py
+src/legacy/configuration.py
+src/legacy/graph.py
+src/legacy/multi_agent.py
+src/legacy/prompts.py
+src/legacy/state.py
+src/legacy/utils.py
+src/open_deep_research/configuration.py
+src/open_deep_research/deep_researcher.py
+src/open_deep_research/prompts.py
+src/open_deep_research/state.py
+src/open_deep_research/utils.py
+src/open_research_lite/__init__.py
+tests/evaluators.py
+tests/extract_langsmith_data.py
+tests/pairwise_evaluation.py
+tests/prompts.py
+tests/run_evaluate.py
+tests/supervisor_parallel_evaluation.py
+tests/test_concept_diff.py

open_research_lite-0.1.0/open_research_lite.egg-info/dependency_links.txt

@@ -0,0 +1 @@
+

open_research_lite-0.1.0/open_research_lite.egg-info/requires.txt

@@ -0,0 +1,18 @@
+langgraph>=0.5.4
+langchain-community>=0.3.9
+langchain-openai>=0.3.28
+langchain-anthropic>=0.3.15
+langchain-mcp-adapters>=0.1.6
+langchain-tavily
+openai>=1.99.2
+tavily-python>=0.5.0
+pydantic>=2.0.0
+requests>=2.32.3
+beautifulsoup4==4.14.3
+python-dotenv>=1.0.1
+pytest
+langchain-google-genai>=2.1.5
+
+[dev]
+mypy>=1.11.1
+ruff>=0.6.1

open_research_lite-0.1.0/open_research_lite.egg-info/top_level.txt

@@ -0,0 +1,4 @@
+legacy
+open_deep_research
+open_research_lite
+tests

open_research_lite-0.1.0/pyproject.toml

@@ -0,0 +1,42 @@
+[project]
+name = "open-research-lite"
+version = "0.1.0"
+description = "Token-Efficient Ingestion Middleware & Concept-Diff Engine for Deep Research AI Agents."
+authors = [
+    { name = "DeepResearch-Lite Team" }
+]
+readme = "README.md"
+license = { text = "MIT" }
+requires-python = ">=3.10"
+dependencies = [
+    "langgraph>=0.5.4",
+    "langchain-community>=0.3.9",
+    "langchain-openai>=0.3.28",
+    "langchain-anthropic>=0.3.15",
+    "langchain-mcp-adapters>=0.1.6",
+    "langchain-tavily",
+    "openai>=1.99.2",
+    "tavily-python>=0.5.0",
+    "pydantic>=2.0.0",
+    "requests>=2.32.3",
+    "beautifulsoup4==4.14.3",
+    "python-dotenv>=1.0.1",
+    "pytest",
+    "langchain-google-genai>=2.1.5",
+]
+
+[project.optional-dependencies]
+dev = ["mypy>=1.11.1", "ruff>=0.6.1"]
+
+[build-system]
+requires = ["setuptools>=73.0.0", "wheel"]
+build-backend = "setuptools.build_meta"
+
+[tool.setuptools]
+packages = ["open_research_lite", "open_deep_research", "legacy", "tests"]
+
+[tool.setuptools.package-dir]
+"open_research_lite" = "src/open_research_lite"
+"open_deep_research" = "src/open_deep_research"
+"legacy" = "src/legacy"
+"tests" = "tests"

open_research_lite-0.1.0/setup.cfg

@@ -0,0 +1,4 @@
+[egg_info]
+tag_build = 
+tag_date = 0
+

open_research_lite-0.1.0/src/legacy/__init__.py

@@ -0,0 +1,3 @@
+"""Planning, research, and report generation."""
+
+__version__ = "0.0.15"

open_research_lite-0.1.0/src/legacy/configuration.py

@@ -0,0 +1,106 @@
+import os
+from enum import Enum
+from dataclasses import dataclass, fields
+from typing import Any, Optional, Dict, Literal
+
+from langchain_core.runnables import RunnableConfig
+
+DEFAULT_REPORT_STRUCTURE = """Use this structure to create a report on the user-provided topic:
+
+1. Introduction (no research needed)
+   - Brief overview of the topic area
+
+2. Main Body Sections:
+   - Each section should focus on a sub-topic of the user-provided topic
+   
+3. Conclusion
+   - Aim for 1 structural element (either a list or table) that distills the main body sections 
+   - Provide a concise summary of the report"""
+
+class SearchAPI(Enum):
+    PERPLEXITY = "perplexity"
+    TAVILY = "tavily"
+    EXA = "exa"
+    ARXIV = "arxiv"
+    PUBMED = "pubmed"
+    LINKUP = "linkup"
+    DUCKDUCKGO = "duckduckgo"
+    GOOGLESEARCH = "googlesearch"
+    NONE = "none"
+
+@dataclass(kw_only=True)
+class Configuration:
+    """Configuration for the workflow/graph-based implementation (graph.py)."""
+    # Common configuration
+    report_structure: str = DEFAULT_REPORT_STRUCTURE
+    search_api: SearchAPI = SearchAPI.TAVILY
+    search_api_config: Optional[Dict[str, Any]] = None
+    process_search_results: Literal["summarize", "split_and_rerank"] | None = None
+    summarization_model_provider: str = "openai"
+    summarization_model: str = "gpt-4.1"
+    max_structured_output_retries: int = 3
+    include_source_str: bool = False
+    
+    # Workflow-specific configuration
+    number_of_queries: int = 2 # Number of search queries to generate per iteration
+    max_search_depth: int = 2 # Maximum number of reflection + search iterations
+    planner_provider: str = "anthropic"
+    planner_model: str = "claude-3-7-sonnet-latest"
+    planner_model_kwargs: Optional[Dict[str, Any]] = None
+    writer_provider: str = "openai"
+    writer_model: str = "gpt-4.1"
+    writer_model_kwargs: Optional[Dict[str, Any]] = None
+
+    @classmethod
+    def from_runnable_config(
+        cls, config: Optional[RunnableConfig] = None
+    ) -> "Configuration":
+        """Create a Configuration instance from a RunnableConfig."""
+        configurable = (
+            config["configurable"] if config and "configurable" in config else {}
+        )
+        values: dict[str, Any] = {
+            f.name: os.environ.get(f.name.upper(), configurable.get(f.name))
+            for f in fields(cls)
+            if f.init
+        }
+        return cls(**{k: v for k, v in values.items() if v})
+
+@dataclass(kw_only=True)
+class MultiAgentConfiguration:
+    """Configuration for the multi-agent implementation (multi_agent.py)."""
+    # Common configuration
+    search_api: SearchAPI = SearchAPI.TAVILY
+    search_api_config: Optional[Dict[str, Any]] = None
+    process_search_results: Literal["summarize", "split_and_rerank"] | None = None
+    summarization_model_provider: str = "openai"
+    summarization_model: str = "gpt-4.1"
+    include_source_str: bool = False
+    
+    # Multi-agent specific configuration
+    number_of_queries: int = 2 # Number of search queries to generate per section
+    supervisor_model: str = "anthropic:claude-sonnet-4-20250514"
+    researcher_model: str = "anthropic:claude-sonnet-4-20250514"
+    ask_for_clarification: bool = False # Whether to ask for clarification from the user
+    # MCP server configuration
+    mcp_server_config: Optional[Dict[str, Any]] = None
+    mcp_prompt: Optional[str] = None
+    mcp_tools_to_include: Optional[list[str]] = None
+
+    @classmethod
+    def from_runnable_config(
+        cls, config: Optional[RunnableConfig] = None
+    ) -> "MultiAgentConfiguration":
+        """Create a MultiAgentConfiguration instance from a RunnableConfig."""
+        configurable = (
+            config["configurable"] if config and "configurable" in config else {}
+        )
+        values: dict[str, Any] = {
+            f.name: os.environ.get(f.name.upper(), configurable.get(f.name))
+            for f in fields(cls)
+            if f.init
+        }
+        return cls(**{k: v for k, v in values.items() if v})
+
+# Keep the old Configuration class for backward compatibility
+Configuration = Configuration