Repository files navigation

Deep Research Agent

A plugin-based agentic research framework built on LangGraph. It decomposes a research question into sub-questions, retrieves and scrapes information from multiple sources, checks its own work for gaps, and synthesizes a structured, cited report — with every search provider, scraper, and LLM backend swappable through configuration.

PythonLangGraphStatus

Table of Contents

Why This Exists

Most "research agent" scripts hard-code a single search API and a single model. This framework treats every external dependency — search, scraping, and LLM inference — as a pluggable, prioritized, fallback-capable tool, so a provider outage or missing API key degrades gracefully instead of breaking the run. The orchestration itself (planning → retrieval → reflection → synthesis) is a LangGraph state machine that knows nothing about which concrete tools it's calling.

Features

  • Plugin architecture — swap search providers, scrapers, and LLM backends entirely through YAML configuration, no code changes required.
  • Automatic fallback — each tool category (search, scraper, LLM) has a priority-ordered fallback chain; if the top tool fails, the next one is tried automatically.
  • Iterative research loop — the agent plans, retrieves, reflects on what's missing, and loops back to retrieval until the report is complete or max_loops is hit.
  • Structured citations — reports carry inline [n] citations backed by a Citation model with URL, title, excerpt, and access timestamp.
  • Model routing — routes planning/reflection/synthesis to different LLM tiers (fast / balanced / powerful) via LiteLLM, so cheaper models handle cheaper steps.
  • Drop-in custom tools — add a file to tools/custom/, subclass a base tool, and it's auto-discovered on startup.

Architecture

The framework is organized into four layers:

  1. Core orchestration (core/) — a tool-agnostic LangGraph state machine that drives the research loop.
  2. Tool registry (registry/) — discovers, prioritizes, and selects tools, with fallback-chain resolution.
  3. Tool implementations (tools/) — concrete search, scraper, LLM, and custom tool adapters.
  4. Data models (models/) — Pydantic schemas for citations and tool I/O, so every tool boundary is type-checked.

Research Workflow

Query → Planning → Retrieval → Reflection → Synthesis → Report
↑ ↓
└─ (if gaps) ─┘
  1. Planning — decomposes the query into 3–5 sub-questions using a fast LLM.
  2. Retrieval — searches and scrapes content for each sub-question, trying tools down the fallback chain as needed.
  3. Reflection — evaluates completeness against the original query and identifies remaining gaps.
  4. Synthesis — once the reflection step is satisfied (or max_loops is reached), a powerful LLM generates the final cited report.

Installation

Prerequisites

  • Python 3.8+
  • pip (or conda)

Setup

# 1. Clone the repository
git clone https://github.com/QuantumWars/DeepResearchAgent.git
cd DeepResearchAgent
# 2. Install dependencies
pip install -r requirements.txt
# 3. Install Playwright browsers (needed for JS-heavy sites)
playwright install
# 4. Configure API keys
cp .env.example .env

Then edit .env with the providers you plan to use:

# Search (at least one required)
TAVILY_API_KEY=your_tavily_key_here
SERPER_API_KEY=your_serper_key_here
# LLM (at least one required)
OPENAI_API_KEY=your_openai_key_here
# or
ANTHROPIC_API_KEY=your_anthropic_key_here

Quick Start

Command Line

# Basic research query
python main.py "What is quantum computing?"# Save the report to a file
python main.py "Explain climate change" --output report.md
# Cap the number of retrieval/reflection loops
python main.py "AI safety research" --max-loops 2
# Verbose logging for debugging
python main.py "Machine learning basics" --verbose

Programmatic Usage

fromcore.orchestratorimportResearchOrchestratororchestrator=ResearchOrchestrator("config/tool_config.yaml")
result=orchestrator.research(
query="What is quantum computing?",
max_loops=3
)
print(result.report)
print(f"Sources: {len(result.sources)}")
print(f"Tool calls: {len(result.execution_log)}")
result.save("reports/quantum_computing.md")
forcitationinresult.get_citations():
print(f"[{citation.id}] {citation.title} - {citation.url}")

Using Custom Tools

fromcore.orchestratorimportResearchOrchestratorfromregistry.base_toolimportBaseCustomToolclassPDFExtractor(BaseCustomTool):
name="pdf_extractor"description="Extracts text from PDF files"defexecute(self, input_data):
url=input_data.get("url")
# ... extraction logic ...return {"success": True, "content": extracted_text}
orchestrator=ResearchOrchestrator()
result=orchestrator.research(
query="Research question",
custom_tools=[PDFExtractor()]
)

Configuration Guide

All tool selection, priority, and fallback behavior lives in config/tool_config.yaml.

Search Tools

search_tools:
fallback_chain: # tried in order until one succeeds
- tavily
- serpertavily:
enabled: truepriority: 10# higher = tried firstapi_key: env:TAVILY_API_KEYextra_params:
search_depth: basic # or 'advanced'max_results: 5

Available: tavily (fast, AI-optimized — recommended), serper (Google Search API).

Scraper Tools

scraper_tools:
fallback_chain:
- trafilatura
- playwrighttrafilatura:
enabled: truepriority: 10extra_params:
include_tables: truededuplicate: trueplaywright:
enabled: truepriority: 5extra_params:
headless: truetimeout: 30000wait_for: networkidle

Available: trafilatura (fast, lightweight, for standard sites), playwright (full browser automation, for JS-heavy sites).

LLM Tools

llm_tools:
routing:
fast: gpt-3.5-turbo # planningbalanced: gpt-4-turbo-preview # reflectionpowerful: gpt-4 # synthesisprovider: litellmextra_params:
temperature: 0.7max_tokens: 4000

Supported providers (via LiteLLM): OpenAI, Anthropic, Cohere, and anything else LiteLLM speaks.

Workflow Settings

workflow:
max_loops: 3# maximum research iterationsmax_documents: 20# maximum documents to retrievemax_scrape_size: 102400# max bytes per page (100KB)tool_timeout: 30# timeout per tool call (seconds)

Environment Variables

config/tool_config.yaml references secrets with an env:VAR_NAME syntax, resolved from .env:

TAVILY_API_KEY=tvly-xxxxx
SERPER_API_KEY=xxxxx
OPENAI_API_KEY=sk-xxxxx
ANTHROPIC_API_KEY=sk-ant-xxxxx
COHERE_API_KEY=xxxxx

Adding Custom Tools

Drop a file in tools/custom/ and subclass the appropriate base:

# tools/custom/my_tool.pyfromregistry.base_toolimportBaseCustomToolimportlogginglogger=logging.getLogger(__name__)
classMyCustomTool(BaseCustomTool):
"""Description of what your tool does."""name="my_tool"description="Custom tool for specific task"def__init__(self, api_key=None, extra_params=None):
self.api_key=api_keyself.config=extra_paramsor {}
logger.info(f"Initialized {self.name}")
defexecute(self, input_data):
try:
result=self._process(input_data)
return {"success": True, "data": result}
exceptExceptionase:
logger.error(f"{self.name} failed: {e}")
return {"success": False, "error": str(e)}

No manual registration needed — the registry discovers and registers tools on startup. Optionally add config for it under custom_tools: in tool_config.yaml. Pick the right base class:

  • BaseSearchTool — search providers
  • BaseScraperTool — web scrapers
  • BaseLLMTool — LLM integrations
  • BaseCustomTool — everything else

See examples/custom_tool_example.py for a complete walkthrough.

API Reference

ResearchOrchestrator

classResearchOrchestrator:
def__init__(self, config_path: str="config/tool_config.yaml")
defresearch(
self,
query: str,
custom_tools: Optional[List[BaseTool]] =None,
max_loops: int=3
) ->ResearchResult

ResearchResult

classResearchResult:
report: str# markdown report with citationssources: List[dict] # source documentsexecution_log: List[dict] # tool execution historydefsave(self, filepath: str) ->Nonedefget_citations(self) ->List[Citation]

ToolRegistry

classToolRegistry:
defregister_tool(self, tool_instance: BaseTool, category: str, priority: Optional[int] =None) ->Nonedefget_tool(self, category: str, name: Optional[str] =None) ->Optional[BaseTool]
defget_tool_chain(self, category: str) ->List[BaseTool]
@classmethoddeffrom_config(cls, config_path: str) ->"ToolRegistry"defdiscover_tools(self, tools_directory: str="tools") ->int

Project Structure

DeepResearchAgent/
├── core/ # Core orchestration layer
│ ├── orchestrator.py # ResearchOrchestrator / ResearchResult
│ ├── graph.py # LangGraph workflow definition
│ ├── workflow_nodes.py # Planning/retrieval/reflection/synthesis nodes
│ └── state.py # Shared research state
├── registry/ # Tool registry system
│ ├── tool_registry.py # Discovery, priority, fallback chains
│ └── base_tool.py # Abstract base classes
├── tools/ # Tool implementations
│ ├── search/ # tavily_search.py, serper_search.py
│ ├── scraper/ # trafilatura_scraper.py, playwright_scraper.py
│ ├── llm/ # litellm_tool.py
│ └── custom/ # Your custom tools go here
├── models/ # Pydantic schemas
│ └── tool_schemas.py
├── config/
│ └── tool_config.yaml
├── utils/ # Config loading, logging, formatting helpers
├── examples/ # Runnable usage examples
├── main.py # CLI entry point
├── requirements.txt
└── .env.example

Testing

# End-to-end execution smoke test
python test_execution.py
# Usage examples double as integration checks
python examples/basic_research.py
python examples/custom_tool_example.py
python examples/test_orchestrator.py

See TEST_RESULTS.md and test_report.md for the latest recorded run output.

Troubleshooting

Tool 'tavily' failed: API key not found Check that .env exists, contains TAVILY_API_KEY=tvly-xxxxx, and that the process was restarted after adding it.

All search tools failing Verify API key validity, internet connectivity, and rate limits; rerun with --verbose to see per-tool logs.

Playwright scraper timing out Raise timeout under playwright.extra_params in tool_config.yaml, fall back to trafilatura for simpler sites, or check whether the target blocks automation.

ModuleNotFoundError: No module named 'langgraph'pip install -r requirements.txt.

playwright._impl._api_types.Error: Executable doesn't existplaywright install.

Empty report Make the query more specific, confirm at least one search tool is healthy, check the verbose logs, or raise --max-loops.

Research is slow Lower max_loops, reduce max_results on search tools, use faster models in llm_tools.routing, or trim the fallback chains.

Enable verbose logging for any of the above:

python main.py "query" --verbose

Contributing

Contributions are welcome. Areas that could use help:

  • Additional search providers (Brave, Bing, etc.)
  • More scraper implementations
  • New custom node implementations
  • Performance optimizations
  • Documentation improvements

Design docs live under .kiro/specs/deep-research-framework/ if you want the full requirements/design/task breakdown behind the current implementation.

License

No formal license has been selected yet. All rights reserved by QuantumWars until one is chosen — open an issue if you need clarification for a specific use case.

About

Plugin-based LangGraph research agent that plans, retrieves, reflects, and synthesizes cited reports — with swappable search, scraper, and LLM providers and automatic fallback.

Resources

Stars

0 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages

, 'i'); if (__m === '*' || __re.test(location.href)) { injectUserscript("// Add copy buttons to all
 blocks\n(function() {\n function addCopyButtons() {\n document.querySelectorAll('pre code').forEach(function(codeBlock) {\n if (codeBlock.parentElement.hasAttribute('data-copy-added')) return;\n codeBlock.parentElement.setAttribute('data-copy-added', 'true');\n \n var btn = document.createElement('button');\n btn.textContent = 'Copy';\n btn.style.cssText = 'position:absolute;top:4px;right:4px;padding:2px 8px;font-size:11px;background:#4ecdc4;border:none;border-radius:4px;color:#1a1a2e;cursor:pointer;opacity:0.7;transition:opacity 0.2s;';\n btn.onmouseover = function() { this.style.opacity = '1'; };\n btn.onmouseout = function() { this.style.opacity = '0.7'; };\n btn.onclick = function() {\n navigator.clipboard.writeText(codeBlock.textContent).then(function() {\n btn.textContent = 'Copied!';\n setTimeout(function() { btn.textContent = 'Copy'; }, 1500);\n });\n };\n codeBlock.parentElement.style.position = 'relative';\n codeBlock.parentElement.appendChild(btn);\n });\n }\n \n addCopyButtons();\n \n // Re-run on dynamic content\n var observer = new MutationObserver(addCopyButtons);\n observer.observe(document.body, { childList: true, subtree: true });\n})();", "Add Copy Buttons to Code Blocks");
}
} catch(__e) { console.warn('[Userscript:Add Copy Buttons to Code Blocks]', __e); }
})();
(function(){
try {
var __m = "github.com";
var __re = new RegExp('^' + "github\\.com" + '
Skip to content

Repository files navigation

Deep Research Agent

A plugin-based agentic research framework built on LangGraph. It decomposes a research question into sub-questions, retrieves and scrapes information from multiple sources, checks its own work for gaps, and synthesizes a structured, cited report — with every search provider, scraper, and LLM backend swappable through configuration.

PythonLangGraphStatus

Table of Contents

Why This Exists

Most "research agent" scripts hard-code a single search API and a single model. This framework treats every external dependency — search, scraping, and LLM inference — as a pluggable, prioritized, fallback-capable tool, so a provider outage or missing API key degrades gracefully instead of breaking the run. The orchestration itself (planning → retrieval → reflection → synthesis) is a LangGraph state machine that knows nothing about which concrete tools it's calling.

Features

  • Plugin architecture — swap search providers, scrapers, and LLM backends entirely through YAML configuration, no code changes required.
  • Automatic fallback — each tool category (search, scraper, LLM) has a priority-ordered fallback chain; if the top tool fails, the next one is tried automatically.
  • Iterative research loop — the agent plans, retrieves, reflects on what's missing, and loops back to retrieval until the report is complete or max_loops is hit.
  • Structured citations — reports carry inline [n] citations backed by a Citation model with URL, title, excerpt, and access timestamp.
  • Model routing — routes planning/reflection/synthesis to different LLM tiers (fast / balanced / powerful) via LiteLLM, so cheaper models handle cheaper steps.
  • Drop-in custom tools — add a file to tools/custom/, subclass a base tool, and it's auto-discovered on startup.

Architecture

The framework is organized into four layers:

  1. Core orchestration (core/) — a tool-agnostic LangGraph state machine that drives the research loop.
  2. Tool registry (registry/) — discovers, prioritizes, and selects tools, with fallback-chain resolution.
  3. Tool implementations (tools/) — concrete search, scraper, LLM, and custom tool adapters.
  4. Data models (models/) — Pydantic schemas for citations and tool I/O, so every tool boundary is type-checked.

Research Workflow

Query → Planning → Retrieval → Reflection → Synthesis → Report
↑ ↓
└─ (if gaps) ─┘
  1. Planning — decomposes the query into 3–5 sub-questions using a fast LLM.
  2. Retrieval — searches and scrapes content for each sub-question, trying tools down the fallback chain as needed.
  3. Reflection — evaluates completeness against the original query and identifies remaining gaps.
  4. Synthesis — once the reflection step is satisfied (or max_loops is reached), a powerful LLM generates the final cited report.

Installation

Prerequisites

  • Python 3.8+
  • pip (or conda)

Setup

# 1. Clone the repository
git clone https://github.com/QuantumWars/DeepResearchAgent.git
cd DeepResearchAgent
# 2. Install dependencies
pip install -r requirements.txt
# 3. Install Playwright browsers (needed for JS-heavy sites)
playwright install
# 4. Configure API keys
cp .env.example .env

Then edit .env with the providers you plan to use:

# Search (at least one required)
TAVILY_API_KEY=your_tavily_key_here
SERPER_API_KEY=your_serper_key_here
# LLM (at least one required)
OPENAI_API_KEY=your_openai_key_here
# or
ANTHROPIC_API_KEY=your_anthropic_key_here

Quick Start

Command Line

# Basic research query
python main.py "What is quantum computing?"# Save the report to a file
python main.py "Explain climate change" --output report.md
# Cap the number of retrieval/reflection loops
python main.py "AI safety research" --max-loops 2
# Verbose logging for debugging
python main.py "Machine learning basics" --verbose

Programmatic Usage

fromcore.orchestratorimportResearchOrchestratororchestrator=ResearchOrchestrator("config/tool_config.yaml")
result=orchestrator.research(
query="What is quantum computing?",
max_loops=3
)
print(result.report)
print(f"Sources: {len(result.sources)}")
print(f"Tool calls: {len(result.execution_log)}")
result.save("reports/quantum_computing.md")
forcitationinresult.get_citations():
print(f"[{citation.id}] {citation.title} - {citation.url}")

Using Custom Tools

fromcore.orchestratorimportResearchOrchestratorfromregistry.base_toolimportBaseCustomToolclassPDFExtractor(BaseCustomTool):
name="pdf_extractor"description="Extracts text from PDF files"defexecute(self, input_data):
url=input_data.get("url")
# ... extraction logic ...return {"success": True, "content": extracted_text}
orchestrator=ResearchOrchestrator()
result=orchestrator.research(
query="Research question",
custom_tools=[PDFExtractor()]
)

Configuration Guide

All tool selection, priority, and fallback behavior lives in config/tool_config.yaml.

Search Tools

search_tools:
fallback_chain: # tried in order until one succeeds
- tavily
- serpertavily:
enabled: truepriority: 10# higher = tried firstapi_key: env:TAVILY_API_KEYextra_params:
search_depth: basic # or 'advanced'max_results: 5

Available: tavily (fast, AI-optimized — recommended), serper (Google Search API).

Scraper Tools

scraper_tools:
fallback_chain:
- trafilatura
- playwrighttrafilatura:
enabled: truepriority: 10extra_params:
include_tables: truededuplicate: trueplaywright:
enabled: truepriority: 5extra_params:
headless: truetimeout: 30000wait_for: networkidle

Available: trafilatura (fast, lightweight, for standard sites), playwright (full browser automation, for JS-heavy sites).

LLM Tools

llm_tools:
routing:
fast: gpt-3.5-turbo # planningbalanced: gpt-4-turbo-preview # reflectionpowerful: gpt-4 # synthesisprovider: litellmextra_params:
temperature: 0.7max_tokens: 4000

Supported providers (via LiteLLM): OpenAI, Anthropic, Cohere, and anything else LiteLLM speaks.

Workflow Settings

workflow:
max_loops: 3# maximum research iterationsmax_documents: 20# maximum documents to retrievemax_scrape_size: 102400# max bytes per page (100KB)tool_timeout: 30# timeout per tool call (seconds)

Environment Variables

config/tool_config.yaml references secrets with an env:VAR_NAME syntax, resolved from .env:

TAVILY_API_KEY=tvly-xxxxx
SERPER_API_KEY=xxxxx
OPENAI_API_KEY=sk-xxxxx
ANTHROPIC_API_KEY=sk-ant-xxxxx
COHERE_API_KEY=xxxxx

Adding Custom Tools

Drop a file in tools/custom/ and subclass the appropriate base:

# tools/custom/my_tool.pyfromregistry.base_toolimportBaseCustomToolimportlogginglogger=logging.getLogger(__name__)
classMyCustomTool(BaseCustomTool):
"""Description of what your tool does."""name="my_tool"description="Custom tool for specific task"def__init__(self, api_key=None, extra_params=None):
self.api_key=api_keyself.config=extra_paramsor {}
logger.info(f"Initialized {self.name}")
defexecute(self, input_data):
try:
result=self._process(input_data)
return {"success": True, "data": result}
exceptExceptionase:
logger.error(f"{self.name} failed: {e}")
return {"success": False, "error": str(e)}

No manual registration needed — the registry discovers and registers tools on startup. Optionally add config for it under custom_tools: in tool_config.yaml. Pick the right base class:

  • BaseSearchTool — search providers
  • BaseScraperTool — web scrapers
  • BaseLLMTool — LLM integrations
  • BaseCustomTool — everything else

See examples/custom_tool_example.py for a complete walkthrough.

API Reference

ResearchOrchestrator

classResearchOrchestrator:
def__init__(self, config_path: str="config/tool_config.yaml")
defresearch(
self,
query: str,
custom_tools: Optional[List[BaseTool]] =None,
max_loops: int=3
) ->ResearchResult

ResearchResult

classResearchResult:
report: str# markdown report with citationssources: List[dict] # source documentsexecution_log: List[dict] # tool execution historydefsave(self, filepath: str) ->Nonedefget_citations(self) ->List[Citation]

ToolRegistry

classToolRegistry:
defregister_tool(self, tool_instance: BaseTool, category: str, priority: Optional[int] =None) ->Nonedefget_tool(self, category: str, name: Optional[str] =None) ->Optional[BaseTool]
defget_tool_chain(self, category: str) ->List[BaseTool]
@classmethoddeffrom_config(cls, config_path: str) ->"ToolRegistry"defdiscover_tools(self, tools_directory: str="tools") ->int

Project Structure

DeepResearchAgent/
├── core/ # Core orchestration layer
│ ├── orchestrator.py # ResearchOrchestrator / ResearchResult
│ ├── graph.py # LangGraph workflow definition
│ ├── workflow_nodes.py # Planning/retrieval/reflection/synthesis nodes
│ └── state.py # Shared research state
├── registry/ # Tool registry system
│ ├── tool_registry.py # Discovery, priority, fallback chains
│ └── base_tool.py # Abstract base classes
├── tools/ # Tool implementations
│ ├── search/ # tavily_search.py, serper_search.py
│ ├── scraper/ # trafilatura_scraper.py, playwright_scraper.py
│ ├── llm/ # litellm_tool.py
│ └── custom/ # Your custom tools go here
├── models/ # Pydantic schemas
│ └── tool_schemas.py
├── config/
│ └── tool_config.yaml
├── utils/ # Config loading, logging, formatting helpers
├── examples/ # Runnable usage examples
├── main.py # CLI entry point
├── requirements.txt
└── .env.example

Testing

# End-to-end execution smoke test
python test_execution.py
# Usage examples double as integration checks
python examples/basic_research.py
python examples/custom_tool_example.py
python examples/test_orchestrator.py

See TEST_RESULTS.md and test_report.md for the latest recorded run output.

Troubleshooting

Tool 'tavily' failed: API key not found Check that .env exists, contains TAVILY_API_KEY=tvly-xxxxx, and that the process was restarted after adding it.

All search tools failing Verify API key validity, internet connectivity, and rate limits; rerun with --verbose to see per-tool logs.

Playwright scraper timing out Raise timeout under playwright.extra_params in tool_config.yaml, fall back to trafilatura for simpler sites, or check whether the target blocks automation.

ModuleNotFoundError: No module named 'langgraph'pip install -r requirements.txt.

playwright._impl._api_types.Error: Executable doesn't existplaywright install.

Empty report Make the query more specific, confirm at least one search tool is healthy, check the verbose logs, or raise --max-loops.

Research is slow Lower max_loops, reduce max_results on search tools, use faster models in llm_tools.routing, or trim the fallback chains.

Enable verbose logging for any of the above:

python main.py "query" --verbose

Contributing

Contributions are welcome. Areas that could use help:

  • Additional search providers (Brave, Bing, etc.)
  • More scraper implementations
  • New custom node implementations
  • Performance optimizations
  • Documentation improvements

Design docs live under .kiro/specs/deep-research-framework/ if you want the full requirements/design/task breakdown behind the current implementation.

License

No formal license has been selected yet. All rights reserved by QuantumWars until one is chosen — open an issue if you need clarification for a specific use case.

About

Plugin-based LangGraph research agent that plans, retrieves, reflects, and synthesizes cited reports — with swappable search, scraper, and LLM providers and automatic fallback.

Resources

Stars

0 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages

, 'i'); if (__m === '*' || __re.test(location.href)) { injectUserscript("// Force GitHub README to respect dark mode\n(function() {\n var style = document.createElement('style');\n style.textContent = '\n .markdown-body {\n color-scheme: dark light;\n }\n .markdown-body pre { background: #161b22 !important; }\n .markdown-body code { background: rgba(110, 118, 129, 0.4) !important; }\n .markdown-body table th, .markdown-body table td { border-color: #30363d !important; }\n .markdown-body img { background: #0d1117; }\n .markdown-body blockquote { border-left-color: #8b949e; }\n .markdown-body hr { border-color: #30363d; }\n ';\n document.head.appendChild(style);\n})();", "GitHub Dark Mode README Fix"); } } catch(__e) { console.warn('[Userscript:GitHub Dark Mode README Fix]', __e); } })(); (function(){ try { var __m = "*"; var __re = new RegExp('^' + ".*" + '
Skip to content

Repository files navigation

Deep Research Agent

A plugin-based agentic research framework built on LangGraph. It decomposes a research question into sub-questions, retrieves and scrapes information from multiple sources, checks its own work for gaps, and synthesizes a structured, cited report — with every search provider, scraper, and LLM backend swappable through configuration.

PythonLangGraphStatus

Table of Contents

Why This Exists

Most "research agent" scripts hard-code a single search API and a single model. This framework treats every external dependency — search, scraping, and LLM inference — as a pluggable, prioritized, fallback-capable tool, so a provider outage or missing API key degrades gracefully instead of breaking the run. The orchestration itself (planning → retrieval → reflection → synthesis) is a LangGraph state machine that knows nothing about which concrete tools it's calling.

Features

  • Plugin architecture — swap search providers, scrapers, and LLM backends entirely through YAML configuration, no code changes required.
  • Automatic fallback — each tool category (search, scraper, LLM) has a priority-ordered fallback chain; if the top tool fails, the next one is tried automatically.
  • Iterative research loop — the agent plans, retrieves, reflects on what's missing, and loops back to retrieval until the report is complete or max_loops is hit.
  • Structured citations — reports carry inline [n] citations backed by a Citation model with URL, title, excerpt, and access timestamp.
  • Model routing — routes planning/reflection/synthesis to different LLM tiers (fast / balanced / powerful) via LiteLLM, so cheaper models handle cheaper steps.
  • Drop-in custom tools — add a file to tools/custom/, subclass a base tool, and it's auto-discovered on startup.

Architecture

The framework is organized into four layers:

  1. Core orchestration (core/) — a tool-agnostic LangGraph state machine that drives the research loop.
  2. Tool registry (registry/) — discovers, prioritizes, and selects tools, with fallback-chain resolution.
  3. Tool implementations (tools/) — concrete search, scraper, LLM, and custom tool adapters.
  4. Data models (models/) — Pydantic schemas for citations and tool I/O, so every tool boundary is type-checked.

Research Workflow

Query → Planning → Retrieval → Reflection → Synthesis → Report
↑ ↓
└─ (if gaps) ─┘
  1. Planning — decomposes the query into 3–5 sub-questions using a fast LLM.
  2. Retrieval — searches and scrapes content for each sub-question, trying tools down the fallback chain as needed.
  3. Reflection — evaluates completeness against the original query and identifies remaining gaps.
  4. Synthesis — once the reflection step is satisfied (or max_loops is reached), a powerful LLM generates the final cited report.

Installation

Prerequisites

  • Python 3.8+
  • pip (or conda)

Setup

# 1. Clone the repository
git clone https://github.com/QuantumWars/DeepResearchAgent.git
cd DeepResearchAgent
# 2. Install dependencies
pip install -r requirements.txt
# 3. Install Playwright browsers (needed for JS-heavy sites)
playwright install
# 4. Configure API keys
cp .env.example .env

Then edit .env with the providers you plan to use:

# Search (at least one required)
TAVILY_API_KEY=your_tavily_key_here
SERPER_API_KEY=your_serper_key_here
# LLM (at least one required)
OPENAI_API_KEY=your_openai_key_here
# or
ANTHROPIC_API_KEY=your_anthropic_key_here

Quick Start

Command Line

# Basic research query
python main.py "What is quantum computing?"# Save the report to a file
python main.py "Explain climate change" --output report.md
# Cap the number of retrieval/reflection loops
python main.py "AI safety research" --max-loops 2
# Verbose logging for debugging
python main.py "Machine learning basics" --verbose

Programmatic Usage

fromcore.orchestratorimportResearchOrchestratororchestrator=ResearchOrchestrator("config/tool_config.yaml")
result=orchestrator.research(
query="What is quantum computing?",
max_loops=3
)
print(result.report)
print(f"Sources: {len(result.sources)}")
print(f"Tool calls: {len(result.execution_log)}")
result.save("reports/quantum_computing.md")
forcitationinresult.get_citations():
print(f"[{citation.id}] {citation.title} - {citation.url}")

Using Custom Tools

fromcore.orchestratorimportResearchOrchestratorfromregistry.base_toolimportBaseCustomToolclassPDFExtractor(BaseCustomTool):
name="pdf_extractor"description="Extracts text from PDF files"defexecute(self, input_data):
url=input_data.get("url")
# ... extraction logic ...return {"success": True, "content": extracted_text}
orchestrator=ResearchOrchestrator()
result=orchestrator.research(
query="Research question",
custom_tools=[PDFExtractor()]
)

Configuration Guide

All tool selection, priority, and fallback behavior lives in config/tool_config.yaml.

Search Tools

search_tools:
fallback_chain: # tried in order until one succeeds
- tavily
- serpertavily:
enabled: truepriority: 10# higher = tried firstapi_key: env:TAVILY_API_KEYextra_params:
search_depth: basic # or 'advanced'max_results: 5

Available: tavily (fast, AI-optimized — recommended), serper (Google Search API).

Scraper Tools

scraper_tools:
fallback_chain:
- trafilatura
- playwrighttrafilatura:
enabled: truepriority: 10extra_params:
include_tables: truededuplicate: trueplaywright:
enabled: truepriority: 5extra_params:
headless: truetimeout: 30000wait_for: networkidle

Available: trafilatura (fast, lightweight, for standard sites), playwright (full browser automation, for JS-heavy sites).

LLM Tools

llm_tools:
routing:
fast: gpt-3.5-turbo # planningbalanced: gpt-4-turbo-preview # reflectionpowerful: gpt-4 # synthesisprovider: litellmextra_params:
temperature: 0.7max_tokens: 4000

Supported providers (via LiteLLM): OpenAI, Anthropic, Cohere, and anything else LiteLLM speaks.

Workflow Settings

workflow:
max_loops: 3# maximum research iterationsmax_documents: 20# maximum documents to retrievemax_scrape_size: 102400# max bytes per page (100KB)tool_timeout: 30# timeout per tool call (seconds)

Environment Variables

config/tool_config.yaml references secrets with an env:VAR_NAME syntax, resolved from .env:

TAVILY_API_KEY=tvly-xxxxx
SERPER_API_KEY=xxxxx
OPENAI_API_KEY=sk-xxxxx
ANTHROPIC_API_KEY=sk-ant-xxxxx
COHERE_API_KEY=xxxxx

Adding Custom Tools

Drop a file in tools/custom/ and subclass the appropriate base:

# tools/custom/my_tool.pyfromregistry.base_toolimportBaseCustomToolimportlogginglogger=logging.getLogger(__name__)
classMyCustomTool(BaseCustomTool):
"""Description of what your tool does."""name="my_tool"description="Custom tool for specific task"def__init__(self, api_key=None, extra_params=None):
self.api_key=api_keyself.config=extra_paramsor {}
logger.info(f"Initialized {self.name}")
defexecute(self, input_data):
try:
result=self._process(input_data)
return {"success": True, "data": result}
exceptExceptionase:
logger.error(f"{self.name} failed: {e}")
return {"success": False, "error": str(e)}

No manual registration needed — the registry discovers and registers tools on startup. Optionally add config for it under custom_tools: in tool_config.yaml. Pick the right base class:

  • BaseSearchTool — search providers
  • BaseScraperTool — web scrapers
  • BaseLLMTool — LLM integrations
  • BaseCustomTool — everything else

See examples/custom_tool_example.py for a complete walkthrough.

API Reference

ResearchOrchestrator

classResearchOrchestrator:
def__init__(self, config_path: str="config/tool_config.yaml")
defresearch(
self,
query: str,
custom_tools: Optional[List[BaseTool]] =None,
max_loops: int=3
) ->ResearchResult

ResearchResult

classResearchResult:
report: str# markdown report with citationssources: List[dict] # source documentsexecution_log: List[dict] # tool execution historydefsave(self, filepath: str) ->Nonedefget_citations(self) ->List[Citation]

ToolRegistry

classToolRegistry:
defregister_tool(self, tool_instance: BaseTool, category: str, priority: Optional[int] =None) ->Nonedefget_tool(self, category: str, name: Optional[str] =None) ->Optional[BaseTool]
defget_tool_chain(self, category: str) ->List[BaseTool]
@classmethoddeffrom_config(cls, config_path: str) ->"ToolRegistry"defdiscover_tools(self, tools_directory: str="tools") ->int

Project Structure

DeepResearchAgent/
├── core/ # Core orchestration layer
│ ├── orchestrator.py # ResearchOrchestrator / ResearchResult
│ ├── graph.py # LangGraph workflow definition
│ ├── workflow_nodes.py # Planning/retrieval/reflection/synthesis nodes
│ └── state.py # Shared research state
├── registry/ # Tool registry system
│ ├── tool_registry.py # Discovery, priority, fallback chains
│ └── base_tool.py # Abstract base classes
├── tools/ # Tool implementations
│ ├── search/ # tavily_search.py, serper_search.py
│ ├── scraper/ # trafilatura_scraper.py, playwright_scraper.py
│ ├── llm/ # litellm_tool.py
│ └── custom/ # Your custom tools go here
├── models/ # Pydantic schemas
│ └── tool_schemas.py
├── config/
│ └── tool_config.yaml
├── utils/ # Config loading, logging, formatting helpers
├── examples/ # Runnable usage examples
├── main.py # CLI entry point
├── requirements.txt
└── .env.example

Testing

# End-to-end execution smoke test
python test_execution.py
# Usage examples double as integration checks
python examples/basic_research.py
python examples/custom_tool_example.py
python examples/test_orchestrator.py

See TEST_RESULTS.md and test_report.md for the latest recorded run output.

Troubleshooting

Tool 'tavily' failed: API key not found Check that .env exists, contains TAVILY_API_KEY=tvly-xxxxx, and that the process was restarted after adding it.

All search tools failing Verify API key validity, internet connectivity, and rate limits; rerun with --verbose to see per-tool logs.

Playwright scraper timing out Raise timeout under playwright.extra_params in tool_config.yaml, fall back to trafilatura for simpler sites, or check whether the target blocks automation.

ModuleNotFoundError: No module named 'langgraph'pip install -r requirements.txt.

playwright._impl._api_types.Error: Executable doesn't existplaywright install.

Empty report Make the query more specific, confirm at least one search tool is healthy, check the verbose logs, or raise --max-loops.

Research is slow Lower max_loops, reduce max_results on search tools, use faster models in llm_tools.routing, or trim the fallback chains.

Enable verbose logging for any of the above:

python main.py "query" --verbose

Contributing

Contributions are welcome. Areas that could use help:

  • Additional search providers (Brave, Bing, etc.)
  • More scraper implementations
  • New custom node implementations
  • Performance optimizations
  • Documentation improvements

Design docs live under .kiro/specs/deep-research-framework/ if you want the full requirements/design/task breakdown behind the current implementation.

License

No formal license has been selected yet. All rights reserved by QuantumWars until one is chosen — open an issue if you need clarification for a specific use case.

About

Plugin-based LangGraph research agent that plans, retrieves, reflects, and synthesizes cited reports — with swappable search, scraper, and LLM providers and automatic fallback.

Resources

Stars

0 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages

, 'i'); if (__m === '*' || __re.test(location.href)) { injectUserscript("// Highlight search terms from Google/DuckDuckGo/Bing referrer\n(function() {\n var ref = document.referrer;\n var terms = [];\n \n if (ref.includes('google.com') || ref.includes('duckduckgo.com') || ref.includes('bing.com')) {\n var url = new URL(ref);\n var q = url.searchParams.get('q') || url.searchParams.get('p');\n if (q) {\n terms = q.split(/\\s+/).filter(function(t) { return t.length > 2; });\n }\n }\n \n if (terms.length === 0) return;\n \n var style = document.createElement('style');\n style.textContent = '.userscript-highlight { background: #fbbf24; color: #1a1a2e; padding: 1px 3px; border-radius: 2px; }';\n document.head.appendChild(style);\n \n function highlight(node) {\n if (node.nodeType === 3) { // text node\n var text = node.textContent;\n var found = false;\n terms.forEach(function(term) {\n var regex = new RegExp('(' + term.replace(/[.*+?^${}()|[\\]\\\\]/g, '\\\\') + ')', 'gi');\n if (regex.test(text)) {\n found = true;\n var frag = document.createDocumentFragment();\n var parts = text.split(regex);\n parts.forEach(function(part, i) {\n if (i % 2 === 0) {\n frag.appendChild(document.createTextNode(part));\n } else {\n var span = document.createElement('span');\n span.className = 'userscript-highlight';\n span.textContent = part;\n frag.appendChild(span);\n }\n });\n node.parentNode.replaceChild(frag, node);\n }\n });\n } else if (node.nodeType === 1 && node.childNodes) { // element\n var skipTags = ['SCRIPT', 'STYLE', 'NOSCRIPT', 'TEXTAREA', 'INPUT', 'SELECT'];\n if (!skipTags.includes(node.tagName)) {\n Array.from(node.childNodes).forEach(highlight);\n }\n }\n }\n \n highlight(document.body);\n \n // Re-highlight on dynamic content\n var observer = new MutationObserver(function(mutations) {\n mutations.forEach(function(m) {\n m.addedNodes.forEach(function(node) {\n if (node.nodeType === 1 || node.nodeType === 3) highlight(node);\n });\n });\n });\n observer.observe(document.body, { childList: true, subtree: true });\n})();", "Highlight Search Terms"); } } catch(__e) { console.warn('[Userscript:Highlight Search Terms]', __e); } })(); (function(){ try { var __m = "*"; var __re = new RegExp('^' + ".*" + '
Skip to content

Repository files navigation

Deep Research Agent

A plugin-based agentic research framework built on LangGraph. It decomposes a research question into sub-questions, retrieves and scrapes information from multiple sources, checks its own work for gaps, and synthesizes a structured, cited report — with every search provider, scraper, and LLM backend swappable through configuration.

PythonLangGraphStatus

Table of Contents

Why This Exists

Most "research agent" scripts hard-code a single search API and a single model. This framework treats every external dependency — search, scraping, and LLM inference — as a pluggable, prioritized, fallback-capable tool, so a provider outage or missing API key degrades gracefully instead of breaking the run. The orchestration itself (planning → retrieval → reflection → synthesis) is a LangGraph state machine that knows nothing about which concrete tools it's calling.

Features

  • Plugin architecture — swap search providers, scrapers, and LLM backends entirely through YAML configuration, no code changes required.
  • Automatic fallback — each tool category (search, scraper, LLM) has a priority-ordered fallback chain; if the top tool fails, the next one is tried automatically.
  • Iterative research loop — the agent plans, retrieves, reflects on what's missing, and loops back to retrieval until the report is complete or max_loops is hit.
  • Structured citations — reports carry inline [n] citations backed by a Citation model with URL, title, excerpt, and access timestamp.
  • Model routing — routes planning/reflection/synthesis to different LLM tiers (fast / balanced / powerful) via LiteLLM, so cheaper models handle cheaper steps.
  • Drop-in custom tools — add a file to tools/custom/, subclass a base tool, and it's auto-discovered on startup.

Architecture

The framework is organized into four layers:

  1. Core orchestration (core/) — a tool-agnostic LangGraph state machine that drives the research loop.
  2. Tool registry (registry/) — discovers, prioritizes, and selects tools, with fallback-chain resolution.
  3. Tool implementations (tools/) — concrete search, scraper, LLM, and custom tool adapters.
  4. Data models (models/) — Pydantic schemas for citations and tool I/O, so every tool boundary is type-checked.

Research Workflow

Query → Planning → Retrieval → Reflection → Synthesis → Report
↑ ↓
└─ (if gaps) ─┘
  1. Planning — decomposes the query into 3–5 sub-questions using a fast LLM.
  2. Retrieval — searches and scrapes content for each sub-question, trying tools down the fallback chain as needed.
  3. Reflection — evaluates completeness against the original query and identifies remaining gaps.
  4. Synthesis — once the reflection step is satisfied (or max_loops is reached), a powerful LLM generates the final cited report.

Installation

Prerequisites

  • Python 3.8+
  • pip (or conda)

Setup

# 1. Clone the repository
git clone https://github.com/QuantumWars/DeepResearchAgent.git
cd DeepResearchAgent
# 2. Install dependencies
pip install -r requirements.txt
# 3. Install Playwright browsers (needed for JS-heavy sites)
playwright install
# 4. Configure API keys
cp .env.example .env

Then edit .env with the providers you plan to use:

# Search (at least one required)
TAVILY_API_KEY=your_tavily_key_here
SERPER_API_KEY=your_serper_key_here
# LLM (at least one required)
OPENAI_API_KEY=your_openai_key_here
# or
ANTHROPIC_API_KEY=your_anthropic_key_here

Quick Start

Command Line

# Basic research query
python main.py "What is quantum computing?"# Save the report to a file
python main.py "Explain climate change" --output report.md
# Cap the number of retrieval/reflection loops
python main.py "AI safety research" --max-loops 2
# Verbose logging for debugging
python main.py "Machine learning basics" --verbose

Programmatic Usage

fromcore.orchestratorimportResearchOrchestratororchestrator=ResearchOrchestrator("config/tool_config.yaml")
result=orchestrator.research(
query="What is quantum computing?",
max_loops=3
)
print(result.report)
print(f"Sources: {len(result.sources)}")
print(f"Tool calls: {len(result.execution_log)}")
result.save("reports/quantum_computing.md")
forcitationinresult.get_citations():
print(f"[{citation.id}] {citation.title} - {citation.url}")

Using Custom Tools

fromcore.orchestratorimportResearchOrchestratorfromregistry.base_toolimportBaseCustomToolclassPDFExtractor(BaseCustomTool):
name="pdf_extractor"description="Extracts text from PDF files"defexecute(self, input_data):
url=input_data.get("url")
# ... extraction logic ...return {"success": True, "content": extracted_text}
orchestrator=ResearchOrchestrator()
result=orchestrator.research(
query="Research question",
custom_tools=[PDFExtractor()]
)

Configuration Guide

All tool selection, priority, and fallback behavior lives in config/tool_config.yaml.

Search Tools

search_tools:
fallback_chain: # tried in order until one succeeds
- tavily
- serpertavily:
enabled: truepriority: 10# higher = tried firstapi_key: env:TAVILY_API_KEYextra_params:
search_depth: basic # or 'advanced'max_results: 5

Available: tavily (fast, AI-optimized — recommended), serper (Google Search API).

Scraper Tools

scraper_tools:
fallback_chain:
- trafilatura
- playwrighttrafilatura:
enabled: truepriority: 10extra_params:
include_tables: truededuplicate: trueplaywright:
enabled: truepriority: 5extra_params:
headless: truetimeout: 30000wait_for: networkidle

Available: trafilatura (fast, lightweight, for standard sites), playwright (full browser automation, for JS-heavy sites).

LLM Tools

llm_tools:
routing:
fast: gpt-3.5-turbo # planningbalanced: gpt-4-turbo-preview # reflectionpowerful: gpt-4 # synthesisprovider: litellmextra_params:
temperature: 0.7max_tokens: 4000

Supported providers (via LiteLLM): OpenAI, Anthropic, Cohere, and anything else LiteLLM speaks.

Workflow Settings

workflow:
max_loops: 3# maximum research iterationsmax_documents: 20# maximum documents to retrievemax_scrape_size: 102400# max bytes per page (100KB)tool_timeout: 30# timeout per tool call (seconds)

Environment Variables

config/tool_config.yaml references secrets with an env:VAR_NAME syntax, resolved from .env:

TAVILY_API_KEY=tvly-xxxxx
SERPER_API_KEY=xxxxx
OPENAI_API_KEY=sk-xxxxx
ANTHROPIC_API_KEY=sk-ant-xxxxx
COHERE_API_KEY=xxxxx

Adding Custom Tools

Drop a file in tools/custom/ and subclass the appropriate base:

# tools/custom/my_tool.pyfromregistry.base_toolimportBaseCustomToolimportlogginglogger=logging.getLogger(__name__)
classMyCustomTool(BaseCustomTool):
"""Description of what your tool does."""name="my_tool"description="Custom tool for specific task"def__init__(self, api_key=None, extra_params=None):
self.api_key=api_keyself.config=extra_paramsor {}
logger.info(f"Initialized {self.name}")
defexecute(self, input_data):
try:
result=self._process(input_data)
return {"success": True, "data": result}
exceptExceptionase:
logger.error(f"{self.name} failed: {e}")
return {"success": False, "error": str(e)}

No manual registration needed — the registry discovers and registers tools on startup. Optionally add config for it under custom_tools: in tool_config.yaml. Pick the right base class:

  • BaseSearchTool — search providers
  • BaseScraperTool — web scrapers
  • BaseLLMTool — LLM integrations
  • BaseCustomTool — everything else

See examples/custom_tool_example.py for a complete walkthrough.

API Reference

ResearchOrchestrator

classResearchOrchestrator:
def__init__(self, config_path: str="config/tool_config.yaml")
defresearch(
self,
query: str,
custom_tools: Optional[List[BaseTool]] =None,
max_loops: int=3
) ->ResearchResult

ResearchResult

classResearchResult:
report: str# markdown report with citationssources: List[dict] # source documentsexecution_log: List[dict] # tool execution historydefsave(self, filepath: str) ->Nonedefget_citations(self) ->List[Citation]

ToolRegistry

classToolRegistry:
defregister_tool(self, tool_instance: BaseTool, category: str, priority: Optional[int] =None) ->Nonedefget_tool(self, category: str, name: Optional[str] =None) ->Optional[BaseTool]
defget_tool_chain(self, category: str) ->List[BaseTool]
@classmethoddeffrom_config(cls, config_path: str) ->"ToolRegistry"defdiscover_tools(self, tools_directory: str="tools") ->int

Project Structure

DeepResearchAgent/
├── core/ # Core orchestration layer
│ ├── orchestrator.py # ResearchOrchestrator / ResearchResult
│ ├── graph.py # LangGraph workflow definition
│ ├── workflow_nodes.py # Planning/retrieval/reflection/synthesis nodes
│ └── state.py # Shared research state
├── registry/ # Tool registry system
│ ├── tool_registry.py # Discovery, priority, fallback chains
│ └── base_tool.py # Abstract base classes
├── tools/ # Tool implementations
│ ├── search/ # tavily_search.py, serper_search.py
│ ├── scraper/ # trafilatura_scraper.py, playwright_scraper.py
│ ├── llm/ # litellm_tool.py
│ └── custom/ # Your custom tools go here
├── models/ # Pydantic schemas
│ └── tool_schemas.py
├── config/
│ └── tool_config.yaml
├── utils/ # Config loading, logging, formatting helpers
├── examples/ # Runnable usage examples
├── main.py # CLI entry point
├── requirements.txt
└── .env.example

Testing

# End-to-end execution smoke test
python test_execution.py
# Usage examples double as integration checks
python examples/basic_research.py
python examples/custom_tool_example.py
python examples/test_orchestrator.py

See TEST_RESULTS.md and test_report.md for the latest recorded run output.

Troubleshooting

Tool 'tavily' failed: API key not found Check that .env exists, contains TAVILY_API_KEY=tvly-xxxxx, and that the process was restarted after adding it.

All search tools failing Verify API key validity, internet connectivity, and rate limits; rerun with --verbose to see per-tool logs.

Playwright scraper timing out Raise timeout under playwright.extra_params in tool_config.yaml, fall back to trafilatura for simpler sites, or check whether the target blocks automation.

ModuleNotFoundError: No module named 'langgraph'pip install -r requirements.txt.

playwright._impl._api_types.Error: Executable doesn't existplaywright install.

Empty report Make the query more specific, confirm at least one search tool is healthy, check the verbose logs, or raise --max-loops.

Research is slow Lower max_loops, reduce max_results on search tools, use faster models in llm_tools.routing, or trim the fallback chains.

Enable verbose logging for any of the above:

python main.py "query" --verbose

Contributing

Contributions are welcome. Areas that could use help:

  • Additional search providers (Brave, Bing, etc.)
  • More scraper implementations
  • New custom node implementations
  • Performance optimizations
  • Documentation improvements

Design docs live under .kiro/specs/deep-research-framework/ if you want the full requirements/design/task breakdown behind the current implementation.

License

No formal license has been selected yet. All rights reserved by QuantumWars until one is chosen — open an issue if you need clarification for a specific use case.

About

Plugin-based LangGraph research agent that plans, retrieves, reflects, and synthesizes cited reports — with swappable search, scraper, and LLM providers and automatic fallback.

Resources

Stars

0 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages

, 'i'); if (__m === '*' || __re.test(location.href)) { injectUserscript("// Strip utm_, fbclid, gclid, etc. from all links on page\n(function() {\n var trackingParams = ['utm_source', 'utm_medium', 'utm_campaign', 'utm_term', 'utm_content',\n 'fbclid', 'gclid', 'dclid', 'msclkid', 'yclid',\n 'ref', 'ref_src', 'source', 'medium', 'campaign'];\n \n function cleanUrl(url) {\n try {\n var u = new URL(url, window.location.origin);\n var changed = false;\n trackingParams.forEach(function(p) {\n if (u.searchParams.has(p)) {\n u.searchParams.delete(p);\n changed = true;\n }\n });\n return changed ? u.toString() : url;\n } catch (e) {\n return url;\n }\n }\n \n function cleanLinks() {\n document.querySelectorAll('a[href]').forEach(function(a) {\n var clean = cleanUrl(a.href);\n if (clean !== a.href) a.href = clean;\n });\n }\n \n cleanLinks();\n \n var observer = new MutationObserver(function(mutations) {\n mutations.forEach(function(m) {\n m.addedNodes.forEach(function(node) {\n if (node.nodeType === 1) {\n if (node.tagName === 'A') cleanLinks();\n node.querySelectorAll('a[href]').forEach(function(a) {\n var clean = cleanUrl(a.href);\n if (clean !== a.href) a.href = clean;\n });\n }\n });\n });\n });\n observer.observe(document.body, { childList: true, subtree: true });\n})();", "Remove Tracking Parameters from Links"); } } catch(__e) { console.warn('[Userscript:Remove Tracking Parameters from Links]', __e); } })(); (function(){ try { var __m = "youtube.com"; var __re = new RegExp('^' + "youtube\\.com" + '
Skip to content

Repository files navigation

Deep Research Agent

A plugin-based agentic research framework built on LangGraph. It decomposes a research question into sub-questions, retrieves and scrapes information from multiple sources, checks its own work for gaps, and synthesizes a structured, cited report — with every search provider, scraper, and LLM backend swappable through configuration.

PythonLangGraphStatus

Table of Contents

Why This Exists

Most "research agent" scripts hard-code a single search API and a single model. This framework treats every external dependency — search, scraping, and LLM inference — as a pluggable, prioritized, fallback-capable tool, so a provider outage or missing API key degrades gracefully instead of breaking the run. The orchestration itself (planning → retrieval → reflection → synthesis) is a LangGraph state machine that knows nothing about which concrete tools it's calling.

Features

  • Plugin architecture — swap search providers, scrapers, and LLM backends entirely through YAML configuration, no code changes required.
  • Automatic fallback — each tool category (search, scraper, LLM) has a priority-ordered fallback chain; if the top tool fails, the next one is tried automatically.
  • Iterative research loop — the agent plans, retrieves, reflects on what's missing, and loops back to retrieval until the report is complete or max_loops is hit.
  • Structured citations — reports carry inline [n] citations backed by a Citation model with URL, title, excerpt, and access timestamp.
  • Model routing — routes planning/reflection/synthesis to different LLM tiers (fast / balanced / powerful) via LiteLLM, so cheaper models handle cheaper steps.
  • Drop-in custom tools — add a file to tools/custom/, subclass a base tool, and it's auto-discovered on startup.

Architecture

The framework is organized into four layers:

  1. Core orchestration (core/) — a tool-agnostic LangGraph state machine that drives the research loop.
  2. Tool registry (registry/) — discovers, prioritizes, and selects tools, with fallback-chain resolution.
  3. Tool implementations (tools/) — concrete search, scraper, LLM, and custom tool adapters.
  4. Data models (models/) — Pydantic schemas for citations and tool I/O, so every tool boundary is type-checked.

Research Workflow

Query → Planning → Retrieval → Reflection → Synthesis → Report
↑ ↓
└─ (if gaps) ─┘
  1. Planning — decomposes the query into 3–5 sub-questions using a fast LLM.
  2. Retrieval — searches and scrapes content for each sub-question, trying tools down the fallback chain as needed.
  3. Reflection — evaluates completeness against the original query and identifies remaining gaps.
  4. Synthesis — once the reflection step is satisfied (or max_loops is reached), a powerful LLM generates the final cited report.

Installation

Prerequisites

  • Python 3.8+
  • pip (or conda)

Setup

# 1. Clone the repository
git clone https://github.com/QuantumWars/DeepResearchAgent.git
cd DeepResearchAgent
# 2. Install dependencies
pip install -r requirements.txt
# 3. Install Playwright browsers (needed for JS-heavy sites)
playwright install
# 4. Configure API keys
cp .env.example .env

Then edit .env with the providers you plan to use:

# Search (at least one required)
TAVILY_API_KEY=your_tavily_key_here
SERPER_API_KEY=your_serper_key_here
# LLM (at least one required)
OPENAI_API_KEY=your_openai_key_here
# or
ANTHROPIC_API_KEY=your_anthropic_key_here

Quick Start

Command Line

# Basic research query
python main.py "What is quantum computing?"# Save the report to a file
python main.py "Explain climate change" --output report.md
# Cap the number of retrieval/reflection loops
python main.py "AI safety research" --max-loops 2
# Verbose logging for debugging
python main.py "Machine learning basics" --verbose

Programmatic Usage

fromcore.orchestratorimportResearchOrchestratororchestrator=ResearchOrchestrator("config/tool_config.yaml")
result=orchestrator.research(
query="What is quantum computing?",
max_loops=3
)
print(result.report)
print(f"Sources: {len(result.sources)}")
print(f"Tool calls: {len(result.execution_log)}")
result.save("reports/quantum_computing.md")
forcitationinresult.get_citations():
print(f"[{citation.id}] {citation.title} - {citation.url}")

Using Custom Tools

fromcore.orchestratorimportResearchOrchestratorfromregistry.base_toolimportBaseCustomToolclassPDFExtractor(BaseCustomTool):
name="pdf_extractor"description="Extracts text from PDF files"defexecute(self, input_data):
url=input_data.get("url")
# ... extraction logic ...return {"success": True, "content": extracted_text}
orchestrator=ResearchOrchestrator()
result=orchestrator.research(
query="Research question",
custom_tools=[PDFExtractor()]
)

Configuration Guide

All tool selection, priority, and fallback behavior lives in config/tool_config.yaml.

Search Tools

search_tools:
fallback_chain: # tried in order until one succeeds
- tavily
- serpertavily:
enabled: truepriority: 10# higher = tried firstapi_key: env:TAVILY_API_KEYextra_params:
search_depth: basic # or 'advanced'max_results: 5

Available: tavily (fast, AI-optimized — recommended), serper (Google Search API).

Scraper Tools

scraper_tools:
fallback_chain:
- trafilatura
- playwrighttrafilatura:
enabled: truepriority: 10extra_params:
include_tables: truededuplicate: trueplaywright:
enabled: truepriority: 5extra_params:
headless: truetimeout: 30000wait_for: networkidle

Available: trafilatura (fast, lightweight, for standard sites), playwright (full browser automation, for JS-heavy sites).

LLM Tools

llm_tools:
routing:
fast: gpt-3.5-turbo # planningbalanced: gpt-4-turbo-preview # reflectionpowerful: gpt-4 # synthesisprovider: litellmextra_params:
temperature: 0.7max_tokens: 4000

Supported providers (via LiteLLM): OpenAI, Anthropic, Cohere, and anything else LiteLLM speaks.

Workflow Settings

workflow:
max_loops: 3# maximum research iterationsmax_documents: 20# maximum documents to retrievemax_scrape_size: 102400# max bytes per page (100KB)tool_timeout: 30# timeout per tool call (seconds)

Environment Variables

config/tool_config.yaml references secrets with an env:VAR_NAME syntax, resolved from .env:

TAVILY_API_KEY=tvly-xxxxx
SERPER_API_KEY=xxxxx
OPENAI_API_KEY=sk-xxxxx
ANTHROPIC_API_KEY=sk-ant-xxxxx
COHERE_API_KEY=xxxxx

Adding Custom Tools

Drop a file in tools/custom/ and subclass the appropriate base:

# tools/custom/my_tool.pyfromregistry.base_toolimportBaseCustomToolimportlogginglogger=logging.getLogger(__name__)
classMyCustomTool(BaseCustomTool):
"""Description of what your tool does."""name="my_tool"description="Custom tool for specific task"def__init__(self, api_key=None, extra_params=None):
self.api_key=api_keyself.config=extra_paramsor {}
logger.info(f"Initialized {self.name}")
defexecute(self, input_data):
try:
result=self._process(input_data)
return {"success": True, "data": result}
exceptExceptionase:
logger.error(f"{self.name} failed: {e}")
return {"success": False, "error": str(e)}

No manual registration needed — the registry discovers and registers tools on startup. Optionally add config for it under custom_tools: in tool_config.yaml. Pick the right base class:

  • BaseSearchTool — search providers
  • BaseScraperTool — web scrapers
  • BaseLLMTool — LLM integrations
  • BaseCustomTool — everything else

See examples/custom_tool_example.py for a complete walkthrough.

API Reference

ResearchOrchestrator

classResearchOrchestrator:
def__init__(self, config_path: str="config/tool_config.yaml")
defresearch(
self,
query: str,
custom_tools: Optional[List[BaseTool]] =None,
max_loops: int=3
) ->ResearchResult

ResearchResult

classResearchResult:
report: str# markdown report with citationssources: List[dict] # source documentsexecution_log: List[dict] # tool execution historydefsave(self, filepath: str) ->Nonedefget_citations(self) ->List[Citation]

ToolRegistry

classToolRegistry:
defregister_tool(self, tool_instance: BaseTool, category: str, priority: Optional[int] =None) ->Nonedefget_tool(self, category: str, name: Optional[str] =None) ->Optional[BaseTool]
defget_tool_chain(self, category: str) ->List[BaseTool]
@classmethoddeffrom_config(cls, config_path: str) ->"ToolRegistry"defdiscover_tools(self, tools_directory: str="tools") ->int

Project Structure

DeepResearchAgent/
├── core/ # Core orchestration layer
│ ├── orchestrator.py # ResearchOrchestrator / ResearchResult
│ ├── graph.py # LangGraph workflow definition
│ ├── workflow_nodes.py # Planning/retrieval/reflection/synthesis nodes
│ └── state.py # Shared research state
├── registry/ # Tool registry system
│ ├── tool_registry.py # Discovery, priority, fallback chains
│ └── base_tool.py # Abstract base classes
├── tools/ # Tool implementations
│ ├── search/ # tavily_search.py, serper_search.py
│ ├── scraper/ # trafilatura_scraper.py, playwright_scraper.py
│ ├── llm/ # litellm_tool.py
│ └── custom/ # Your custom tools go here
├── models/ # Pydantic schemas
│ └── tool_schemas.py
├── config/
│ └── tool_config.yaml
├── utils/ # Config loading, logging, formatting helpers
├── examples/ # Runnable usage examples
├── main.py # CLI entry point
├── requirements.txt
└── .env.example

Testing

# End-to-end execution smoke test
python test_execution.py
# Usage examples double as integration checks
python examples/basic_research.py
python examples/custom_tool_example.py
python examples/test_orchestrator.py

See TEST_RESULTS.md and test_report.md for the latest recorded run output.

Troubleshooting

Tool 'tavily' failed: API key not found Check that .env exists, contains TAVILY_API_KEY=tvly-xxxxx, and that the process was restarted after adding it.

All search tools failing Verify API key validity, internet connectivity, and rate limits; rerun with --verbose to see per-tool logs.

Playwright scraper timing out Raise timeout under playwright.extra_params in tool_config.yaml, fall back to trafilatura for simpler sites, or check whether the target blocks automation.

ModuleNotFoundError: No module named 'langgraph'pip install -r requirements.txt.

playwright._impl._api_types.Error: Executable doesn't existplaywright install.

Empty report Make the query more specific, confirm at least one search tool is healthy, check the verbose logs, or raise --max-loops.

Research is slow Lower max_loops, reduce max_results on search tools, use faster models in llm_tools.routing, or trim the fallback chains.

Enable verbose logging for any of the above:

python main.py "query" --verbose

Contributing

Contributions are welcome. Areas that could use help:

  • Additional search providers (Brave, Bing, etc.)
  • More scraper implementations
  • New custom node implementations
  • Performance optimizations
  • Documentation improvements

Design docs live under .kiro/specs/deep-research-framework/ if you want the full requirements/design/task breakdown behind the current implementation.

License

No formal license has been selected yet. All rights reserved by QuantumWars until one is chosen — open an issue if you need clarification for a specific use case.

About

Plugin-based LangGraph research agent that plans, retrieves, reflects, and synthesizes cited reports — with swappable search, scraper, and LLM providers and automatic fallback.

Resources

Stars

0 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages

, 'i'); if (__m === '*' || __re.test(location.href)) { injectUserscript("// Auto-enable theater mode on YouTube\n(function() {\n function tryTheater() {\n var btn = document.querySelector('button[aria-label=\"Theater mode\"], ytd-player #player button[title=\"Theater mode\"]');\n if (btn && !btn.classList.contains('activated')) {\n btn.click();\n }\n }\n \n // Try immediately\n tryTheater();\n \n // Try after navigation (SPA)\n var lastUrl = location.href;\n setInterval(function() {\n if (location.href !== lastUrl) {\n lastUrl = location.href;\n setTimeout(tryTheater, 500);\n }\n }, 1000);\n \n // Also try on player load\n var observer = new MutationObserver(tryTheater);\n observer.observe(document.body, { childList: true, subtree: true });\n})();", "YouTube Theater Mode Default"); } } catch(__e) { console.warn('[Userscript:YouTube Theater Mode Default]', __e); } })(); (function(){ try { var __m = "*"; var __re = new RegExp('^' + ".*" + '
Skip to content

Repository files navigation

Deep Research Agent

A plugin-based agentic research framework built on LangGraph. It decomposes a research question into sub-questions, retrieves and scrapes information from multiple sources, checks its own work for gaps, and synthesizes a structured, cited report — with every search provider, scraper, and LLM backend swappable through configuration.

PythonLangGraphStatus

Table of Contents

Why This Exists

Most "research agent" scripts hard-code a single search API and a single model. This framework treats every external dependency — search, scraping, and LLM inference — as a pluggable, prioritized, fallback-capable tool, so a provider outage or missing API key degrades gracefully instead of breaking the run. The orchestration itself (planning → retrieval → reflection → synthesis) is a LangGraph state machine that knows nothing about which concrete tools it's calling.

Features

  • Plugin architecture — swap search providers, scrapers, and LLM backends entirely through YAML configuration, no code changes required.
  • Automatic fallback — each tool category (search, scraper, LLM) has a priority-ordered fallback chain; if the top tool fails, the next one is tried automatically.
  • Iterative research loop — the agent plans, retrieves, reflects on what's missing, and loops back to retrieval until the report is complete or max_loops is hit.
  • Structured citations — reports carry inline [n] citations backed by a Citation model with URL, title, excerpt, and access timestamp.
  • Model routing — routes planning/reflection/synthesis to different LLM tiers (fast / balanced / powerful) via LiteLLM, so cheaper models handle cheaper steps.
  • Drop-in custom tools — add a file to tools/custom/, subclass a base tool, and it's auto-discovered on startup.

Architecture

The framework is organized into four layers:

  1. Core orchestration (core/) — a tool-agnostic LangGraph state machine that drives the research loop.
  2. Tool registry (registry/) — discovers, prioritizes, and selects tools, with fallback-chain resolution.
  3. Tool implementations (tools/) — concrete search, scraper, LLM, and custom tool adapters.
  4. Data models (models/) — Pydantic schemas for citations and tool I/O, so every tool boundary is type-checked.

Research Workflow

Query → Planning → Retrieval → Reflection → Synthesis → Report
↑ ↓
└─ (if gaps) ─┘
  1. Planning — decomposes the query into 3–5 sub-questions using a fast LLM.
  2. Retrieval — searches and scrapes content for each sub-question, trying tools down the fallback chain as needed.
  3. Reflection — evaluates completeness against the original query and identifies remaining gaps.
  4. Synthesis — once the reflection step is satisfied (or max_loops is reached), a powerful LLM generates the final cited report.

Installation

Prerequisites

  • Python 3.8+
  • pip (or conda)

Setup

# 1. Clone the repository
git clone https://github.com/QuantumWars/DeepResearchAgent.git
cd DeepResearchAgent
# 2. Install dependencies
pip install -r requirements.txt
# 3. Install Playwright browsers (needed for JS-heavy sites)
playwright install
# 4. Configure API keys
cp .env.example .env

Then edit .env with the providers you plan to use:

# Search (at least one required)
TAVILY_API_KEY=your_tavily_key_here
SERPER_API_KEY=your_serper_key_here
# LLM (at least one required)
OPENAI_API_KEY=your_openai_key_here
# or
ANTHROPIC_API_KEY=your_anthropic_key_here

Quick Start

Command Line

# Basic research query
python main.py "What is quantum computing?"# Save the report to a file
python main.py "Explain climate change" --output report.md
# Cap the number of retrieval/reflection loops
python main.py "AI safety research" --max-loops 2
# Verbose logging for debugging
python main.py "Machine learning basics" --verbose

Programmatic Usage

fromcore.orchestratorimportResearchOrchestratororchestrator=ResearchOrchestrator("config/tool_config.yaml")
result=orchestrator.research(
query="What is quantum computing?",
max_loops=3
)
print(result.report)
print(f"Sources: {len(result.sources)}")
print(f"Tool calls: {len(result.execution_log)}")
result.save("reports/quantum_computing.md")
forcitationinresult.get_citations():
print(f"[{citation.id}] {citation.title} - {citation.url}")

Using Custom Tools

fromcore.orchestratorimportResearchOrchestratorfromregistry.base_toolimportBaseCustomToolclassPDFExtractor(BaseCustomTool):
name="pdf_extractor"description="Extracts text from PDF files"defexecute(self, input_data):
url=input_data.get("url")
# ... extraction logic ...return {"success": True, "content": extracted_text}
orchestrator=ResearchOrchestrator()
result=orchestrator.research(
query="Research question",
custom_tools=[PDFExtractor()]
)

Configuration Guide

All tool selection, priority, and fallback behavior lives in config/tool_config.yaml.

Search Tools

search_tools:
fallback_chain: # tried in order until one succeeds
- tavily
- serpertavily:
enabled: truepriority: 10# higher = tried firstapi_key: env:TAVILY_API_KEYextra_params:
search_depth: basic # or 'advanced'max_results: 5

Available: tavily (fast, AI-optimized — recommended), serper (Google Search API).

Scraper Tools

scraper_tools:
fallback_chain:
- trafilatura
- playwrighttrafilatura:
enabled: truepriority: 10extra_params:
include_tables: truededuplicate: trueplaywright:
enabled: truepriority: 5extra_params:
headless: truetimeout: 30000wait_for: networkidle

Available: trafilatura (fast, lightweight, for standard sites), playwright (full browser automation, for JS-heavy sites).

LLM Tools

llm_tools:
routing:
fast: gpt-3.5-turbo # planningbalanced: gpt-4-turbo-preview # reflectionpowerful: gpt-4 # synthesisprovider: litellmextra_params:
temperature: 0.7max_tokens: 4000

Supported providers (via LiteLLM): OpenAI, Anthropic, Cohere, and anything else LiteLLM speaks.

Workflow Settings

workflow:
max_loops: 3# maximum research iterationsmax_documents: 20# maximum documents to retrievemax_scrape_size: 102400# max bytes per page (100KB)tool_timeout: 30# timeout per tool call (seconds)

Environment Variables

config/tool_config.yaml references secrets with an env:VAR_NAME syntax, resolved from .env:

TAVILY_API_KEY=tvly-xxxxx
SERPER_API_KEY=xxxxx
OPENAI_API_KEY=sk-xxxxx
ANTHROPIC_API_KEY=sk-ant-xxxxx
COHERE_API_KEY=xxxxx

Adding Custom Tools

Drop a file in tools/custom/ and subclass the appropriate base:

# tools/custom/my_tool.pyfromregistry.base_toolimportBaseCustomToolimportlogginglogger=logging.getLogger(__name__)
classMyCustomTool(BaseCustomTool):
"""Description of what your tool does."""name="my_tool"description="Custom tool for specific task"def__init__(self, api_key=None, extra_params=None):
self.api_key=api_keyself.config=extra_paramsor {}
logger.info(f"Initialized {self.name}")
defexecute(self, input_data):
try:
result=self._process(input_data)
return {"success": True, "data": result}
exceptExceptionase:
logger.error(f"{self.name} failed: {e}")
return {"success": False, "error": str(e)}

No manual registration needed — the registry discovers and registers tools on startup. Optionally add config for it under custom_tools: in tool_config.yaml. Pick the right base class:

  • BaseSearchTool — search providers
  • BaseScraperTool — web scrapers
  • BaseLLMTool — LLM integrations
  • BaseCustomTool — everything else

See examples/custom_tool_example.py for a complete walkthrough.

API Reference

ResearchOrchestrator

classResearchOrchestrator:
def__init__(self, config_path: str="config/tool_config.yaml")
defresearch(
self,
query: str,
custom_tools: Optional[List[BaseTool]] =None,
max_loops: int=3
) ->ResearchResult

ResearchResult

classResearchResult:
report: str# markdown report with citationssources: List[dict] # source documentsexecution_log: List[dict] # tool execution historydefsave(self, filepath: str) ->Nonedefget_citations(self) ->List[Citation]

ToolRegistry

classToolRegistry:
defregister_tool(self, tool_instance: BaseTool, category: str, priority: Optional[int] =None) ->Nonedefget_tool(self, category: str, name: Optional[str] =None) ->Optional[BaseTool]
defget_tool_chain(self, category: str) ->List[BaseTool]
@classmethoddeffrom_config(cls, config_path: str) ->"ToolRegistry"defdiscover_tools(self, tools_directory: str="tools") ->int

Project Structure

DeepResearchAgent/
├── core/ # Core orchestration layer
│ ├── orchestrator.py # ResearchOrchestrator / ResearchResult
│ ├── graph.py # LangGraph workflow definition
│ ├── workflow_nodes.py # Planning/retrieval/reflection/synthesis nodes
│ └── state.py # Shared research state
├── registry/ # Tool registry system
│ ├── tool_registry.py # Discovery, priority, fallback chains
│ └── base_tool.py # Abstract base classes
├── tools/ # Tool implementations
│ ├── search/ # tavily_search.py, serper_search.py
│ ├── scraper/ # trafilatura_scraper.py, playwright_scraper.py
│ ├── llm/ # litellm_tool.py
│ └── custom/ # Your custom tools go here
├── models/ # Pydantic schemas
│ └── tool_schemas.py
├── config/
│ └── tool_config.yaml
├── utils/ # Config loading, logging, formatting helpers
├── examples/ # Runnable usage examples
├── main.py # CLI entry point
├── requirements.txt
└── .env.example

Testing

# End-to-end execution smoke test
python test_execution.py
# Usage examples double as integration checks
python examples/basic_research.py
python examples/custom_tool_example.py
python examples/test_orchestrator.py

See TEST_RESULTS.md and test_report.md for the latest recorded run output.

Troubleshooting

Tool 'tavily' failed: API key not found Check that .env exists, contains TAVILY_API_KEY=tvly-xxxxx, and that the process was restarted after adding it.

All search tools failing Verify API key validity, internet connectivity, and rate limits; rerun with --verbose to see per-tool logs.

Playwright scraper timing out Raise timeout under playwright.extra_params in tool_config.yaml, fall back to trafilatura for simpler sites, or check whether the target blocks automation.

ModuleNotFoundError: No module named 'langgraph'pip install -r requirements.txt.

playwright._impl._api_types.Error: Executable doesn't existplaywright install.

Empty report Make the query more specific, confirm at least one search tool is healthy, check the verbose logs, or raise --max-loops.

Research is slow Lower max_loops, reduce max_results on search tools, use faster models in llm_tools.routing, or trim the fallback chains.

Enable verbose logging for any of the above:

python main.py "query" --verbose

Contributing

Contributions are welcome. Areas that could use help:

  • Additional search providers (Brave, Bing, etc.)
  • More scraper implementations
  • New custom node implementations
  • Performance optimizations
  • Documentation improvements

Design docs live under .kiro/specs/deep-research-framework/ if you want the full requirements/design/task breakdown behind the current implementation.

License

No formal license has been selected yet. All rights reserved by QuantumWars until one is chosen — open an issue if you need clarification for a specific use case.

About

Plugin-based LangGraph research agent that plans, retrieves, reflects, and synthesizes cited reports — with swappable search, scraper, and LLM providers and automatic fallback.

Resources

Stars

0 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages

, 'i'); if (__m === '*' || __re.test(location.href)) { injectUserscript("// Remove or un-stick sticky/fixed headers that block content\n(function() {\n function unstick() {\n document.querySelectorAll('header, nav, [role=\"banner\"], .header, .navbar, .sticky, .fixed-top, [style*=\"position: fixed\"], [style*=\"position:sticky\"]').forEach(function(el) {\n if (el.style.position === 'fixed' || el.style.position === 'sticky' || \n getComputedStyle(el).position === 'fixed' || getComputedStyle(el).position === 'sticky') {\n el.style.position = 'static';\n el.style.top = 'auto';\n el.style.zIndex = 'auto';\n }\n });\n }\n \n unstick();\n \n var observer = new MutationObserver(unstick);\n observer.observe(document.body, { childList: true, subtree: true, attributes: true, attributeFilter: ['style', 'class'] });\n})();", "Kill Sticky Headers"); } } catch(__e) { console.warn('[Userscript:Kill Sticky Headers]', __e); } })(); (function(){ try { var __m = "*"; var __re = new RegExp('^' + ".*" + '
Skip to content

Repository files navigation

Deep Research Agent

A plugin-based agentic research framework built on LangGraph. It decomposes a research question into sub-questions, retrieves and scrapes information from multiple sources, checks its own work for gaps, and synthesizes a structured, cited report — with every search provider, scraper, and LLM backend swappable through configuration.

PythonLangGraphStatus

Table of Contents

Why This Exists

Most "research agent" scripts hard-code a single search API and a single model. This framework treats every external dependency — search, scraping, and LLM inference — as a pluggable, prioritized, fallback-capable tool, so a provider outage or missing API key degrades gracefully instead of breaking the run. The orchestration itself (planning → retrieval → reflection → synthesis) is a LangGraph state machine that knows nothing about which concrete tools it's calling.

Features

  • Plugin architecture — swap search providers, scrapers, and LLM backends entirely through YAML configuration, no code changes required.
  • Automatic fallback — each tool category (search, scraper, LLM) has a priority-ordered fallback chain; if the top tool fails, the next one is tried automatically.
  • Iterative research loop — the agent plans, retrieves, reflects on what's missing, and loops back to retrieval until the report is complete or max_loops is hit.
  • Structured citations — reports carry inline [n] citations backed by a Citation model with URL, title, excerpt, and access timestamp.
  • Model routing — routes planning/reflection/synthesis to different LLM tiers (fast / balanced / powerful) via LiteLLM, so cheaper models handle cheaper steps.
  • Drop-in custom tools — add a file to tools/custom/, subclass a base tool, and it's auto-discovered on startup.

Architecture

The framework is organized into four layers:

  1. Core orchestration (core/) — a tool-agnostic LangGraph state machine that drives the research loop.
  2. Tool registry (registry/) — discovers, prioritizes, and selects tools, with fallback-chain resolution.
  3. Tool implementations (tools/) — concrete search, scraper, LLM, and custom tool adapters.
  4. Data models (models/) — Pydantic schemas for citations and tool I/O, so every tool boundary is type-checked.

Research Workflow

Query → Planning → Retrieval → Reflection → Synthesis → Report
↑ ↓
└─ (if gaps) ─┘
  1. Planning — decomposes the query into 3–5 sub-questions using a fast LLM.
  2. Retrieval — searches and scrapes content for each sub-question, trying tools down the fallback chain as needed.
  3. Reflection — evaluates completeness against the original query and identifies remaining gaps.
  4. Synthesis — once the reflection step is satisfied (or max_loops is reached), a powerful LLM generates the final cited report.

Installation

Prerequisites

  • Python 3.8+
  • pip (or conda)

Setup

# 1. Clone the repository
git clone https://github.com/QuantumWars/DeepResearchAgent.git
cd DeepResearchAgent
# 2. Install dependencies
pip install -r requirements.txt
# 3. Install Playwright browsers (needed for JS-heavy sites)
playwright install
# 4. Configure API keys
cp .env.example .env

Then edit .env with the providers you plan to use:

# Search (at least one required)
TAVILY_API_KEY=your_tavily_key_here
SERPER_API_KEY=your_serper_key_here
# LLM (at least one required)
OPENAI_API_KEY=your_openai_key_here
# or
ANTHROPIC_API_KEY=your_anthropic_key_here

Quick Start

Command Line

# Basic research query
python main.py "What is quantum computing?"# Save the report to a file
python main.py "Explain climate change" --output report.md
# Cap the number of retrieval/reflection loops
python main.py "AI safety research" --max-loops 2
# Verbose logging for debugging
python main.py "Machine learning basics" --verbose

Programmatic Usage

fromcore.orchestratorimportResearchOrchestratororchestrator=ResearchOrchestrator("config/tool_config.yaml")
result=orchestrator.research(
query="What is quantum computing?",
max_loops=3
)
print(result.report)
print(f"Sources: {len(result.sources)}")
print(f"Tool calls: {len(result.execution_log)}")
result.save("reports/quantum_computing.md")
forcitationinresult.get_citations():
print(f"[{citation.id}] {citation.title} - {citation.url}")

Using Custom Tools

fromcore.orchestratorimportResearchOrchestratorfromregistry.base_toolimportBaseCustomToolclassPDFExtractor(BaseCustomTool):
name="pdf_extractor"description="Extracts text from PDF files"defexecute(self, input_data):
url=input_data.get("url")
# ... extraction logic ...return {"success": True, "content": extracted_text}
orchestrator=ResearchOrchestrator()
result=orchestrator.research(
query="Research question",
custom_tools=[PDFExtractor()]
)

Configuration Guide

All tool selection, priority, and fallback behavior lives in config/tool_config.yaml.

Search Tools

search_tools:
fallback_chain: # tried in order until one succeeds
- tavily
- serpertavily:
enabled: truepriority: 10# higher = tried firstapi_key: env:TAVILY_API_KEYextra_params:
search_depth: basic # or 'advanced'max_results: 5

Available: tavily (fast, AI-optimized — recommended), serper (Google Search API).

Scraper Tools

scraper_tools:
fallback_chain:
- trafilatura
- playwrighttrafilatura:
enabled: truepriority: 10extra_params:
include_tables: truededuplicate: trueplaywright:
enabled: truepriority: 5extra_params:
headless: truetimeout: 30000wait_for: networkidle

Available: trafilatura (fast, lightweight, for standard sites), playwright (full browser automation, for JS-heavy sites).

LLM Tools

llm_tools:
routing:
fast: gpt-3.5-turbo # planningbalanced: gpt-4-turbo-preview # reflectionpowerful: gpt-4 # synthesisprovider: litellmextra_params:
temperature: 0.7max_tokens: 4000

Supported providers (via LiteLLM): OpenAI, Anthropic, Cohere, and anything else LiteLLM speaks.

Workflow Settings

workflow:
max_loops: 3# maximum research iterationsmax_documents: 20# maximum documents to retrievemax_scrape_size: 102400# max bytes per page (100KB)tool_timeout: 30# timeout per tool call (seconds)

Environment Variables

config/tool_config.yaml references secrets with an env:VAR_NAME syntax, resolved from .env:

TAVILY_API_KEY=tvly-xxxxx
SERPER_API_KEY=xxxxx
OPENAI_API_KEY=sk-xxxxx
ANTHROPIC_API_KEY=sk-ant-xxxxx
COHERE_API_KEY=xxxxx

Adding Custom Tools

Drop a file in tools/custom/ and subclass the appropriate base:

# tools/custom/my_tool.pyfromregistry.base_toolimportBaseCustomToolimportlogginglogger=logging.getLogger(__name__)
classMyCustomTool(BaseCustomTool):
"""Description of what your tool does."""name="my_tool"description="Custom tool for specific task"def__init__(self, api_key=None, extra_params=None):
self.api_key=api_keyself.config=extra_paramsor {}
logger.info(f"Initialized {self.name}")
defexecute(self, input_data):
try:
result=self._process(input_data)
return {"success": True, "data": result}
exceptExceptionase:
logger.error(f"{self.name} failed: {e}")
return {"success": False, "error": str(e)}

No manual registration needed — the registry discovers and registers tools on startup. Optionally add config for it under custom_tools: in tool_config.yaml. Pick the right base class:

  • BaseSearchTool — search providers
  • BaseScraperTool — web scrapers
  • BaseLLMTool — LLM integrations
  • BaseCustomTool — everything else

See examples/custom_tool_example.py for a complete walkthrough.

API Reference

ResearchOrchestrator

classResearchOrchestrator:
def__init__(self, config_path: str="config/tool_config.yaml")
defresearch(
self,
query: str,
custom_tools: Optional[List[BaseTool]] =None,
max_loops: int=3
) ->ResearchResult

ResearchResult

classResearchResult:
report: str# markdown report with citationssources: List[dict] # source documentsexecution_log: List[dict] # tool execution historydefsave(self, filepath: str) ->Nonedefget_citations(self) ->List[Citation]

ToolRegistry

classToolRegistry:
defregister_tool(self, tool_instance: BaseTool, category: str, priority: Optional[int] =None) ->Nonedefget_tool(self, category: str, name: Optional[str] =None) ->Optional[BaseTool]
defget_tool_chain(self, category: str) ->List[BaseTool]
@classmethoddeffrom_config(cls, config_path: str) ->"ToolRegistry"defdiscover_tools(self, tools_directory: str="tools") ->int

Project Structure

DeepResearchAgent/
├── core/ # Core orchestration layer
│ ├── orchestrator.py # ResearchOrchestrator / ResearchResult
│ ├── graph.py # LangGraph workflow definition
│ ├── workflow_nodes.py # Planning/retrieval/reflection/synthesis nodes
│ └── state.py # Shared research state
├── registry/ # Tool registry system
│ ├── tool_registry.py # Discovery, priority, fallback chains
│ └── base_tool.py # Abstract base classes
├── tools/ # Tool implementations
│ ├── search/ # tavily_search.py, serper_search.py
│ ├── scraper/ # trafilatura_scraper.py, playwright_scraper.py
│ ├── llm/ # litellm_tool.py
│ └── custom/ # Your custom tools go here
├── models/ # Pydantic schemas
│ └── tool_schemas.py
├── config/
│ └── tool_config.yaml
├── utils/ # Config loading, logging, formatting helpers
├── examples/ # Runnable usage examples
├── main.py # CLI entry point
├── requirements.txt
└── .env.example

Testing

# End-to-end execution smoke test
python test_execution.py
# Usage examples double as integration checks
python examples/basic_research.py
python examples/custom_tool_example.py
python examples/test_orchestrator.py

See TEST_RESULTS.md and test_report.md for the latest recorded run output.

Troubleshooting

Tool 'tavily' failed: API key not found Check that .env exists, contains TAVILY_API_KEY=tvly-xxxxx, and that the process was restarted after adding it.

All search tools failing Verify API key validity, internet connectivity, and rate limits; rerun with --verbose to see per-tool logs.

Playwright scraper timing out Raise timeout under playwright.extra_params in tool_config.yaml, fall back to trafilatura for simpler sites, or check whether the target blocks automation.

ModuleNotFoundError: No module named 'langgraph'pip install -r requirements.txt.

playwright._impl._api_types.Error: Executable doesn't existplaywright install.

Empty report Make the query more specific, confirm at least one search tool is healthy, check the verbose logs, or raise --max-loops.

Research is slow Lower max_loops, reduce max_results on search tools, use faster models in llm_tools.routing, or trim the fallback chains.

Enable verbose logging for any of the above:

python main.py "query" --verbose

Contributing

Contributions are welcome. Areas that could use help:

  • Additional search providers (Brave, Bing, etc.)
  • More scraper implementations
  • New custom node implementations
  • Performance optimizations
  • Documentation improvements

Design docs live under .kiro/specs/deep-research-framework/ if you want the full requirements/design/task breakdown behind the current implementation.

License

No formal license has been selected yet. All rights reserved by QuantumWars until one is chosen — open an issue if you need clarification for a specific use case.

About

Plugin-based LangGraph research agent that plans, retrieves, reflects, and synthesizes cited reports — with swappable search, scraper, and LLM providers and automatic fallback.

Resources

Stars

0 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages

, 'i'); if (__m === '*' || __re.test(location.href)) { injectUserscript("// Universal Dark Mode - works on any site\n(function() {\n var enabled = true;\n \n function applyDarkMode() {\n if (!enabled) return;\n \n // Create style element if it doesn't exist\n var style = document.getElementById('universal-dark-mode-style');\n if (!style) {\n style = document.createElement('style');\n style.id = 'universal-dark-mode-style';\n document.head.appendChild(style);\n }\n \n // Dark mode CSS - inverts colors but preserves images/video\n style.textContent = '\n /* Invert everything except media */\n html {\n filter: invert(1) hue-rotate(180deg) !important;\n background: #1a1a2e !important;\n }\n \n /* Restore images, videos, iframes, canvas */\n img, video, iframe, canvas, svg, picture, [style*=\"background-image\"] {\n filter: invert(1) hue-rotate(180deg) !important;\n }\n \n /* Preserve specific elements that should not be inverted */\n .no-dark-mode, .no-dark-mode *,\n [data-theme=\"light\"], [data-theme=\"light\"],\n .ace_editor, .ace_editor *,\n .CodeMirror, .CodeMirror *,\n .monaco-editor, .monaco-editor *,\n .markdown-body pre, .markdown-body pre *,\n .highlight, .highlight *,\n pre code, pre code * {\n filter: none !important;\n }\n \n /* Fix common UI elements */\n .modal, .popup, .dropdown-menu, .tooltip, .popover {\n filter: invert(1) hue-rotate(180deg) !important;\n background: #2d2d44 !important;\n border-color: #444 !important;\n }\n \n /* Scrollbars */\n ::-webkit-scrollbar { background: #1a1a2e !important; }\n ::-webkit-scrollbar-thumb { background: #444 !important; }\n ::-webkit-scrollbar-thumb:hover { background: #555 !important; }\n \n /* Selection */\n ::selection { background: #4ecdc4 !important; color: #1a1a2e !important; }\n ::-moz-selection { background: #4ecdc4 !important; color: #1a1a2e !important; }\n ';\n }\n \n function removeDarkMode() {\n var style = document.getElementById('universal-dark-mode-style');\n if (style) style.remove();\n }\n \n // Toggle with Alt+Shift+D\n document.addEventListener('keydown', function(e) {\n if (e.altKey && e.shiftKey && e.key === 'D') {\n e.preventDefault();\n enabled = !enabled;\n if (enabled) {\n applyDarkMode();\n console.log('[Universal Dark Mode] Enabled');\n } else {\n removeDarkMode();\n console.log('[Universal Dark Mode] Disabled');\n }\n }\n });\n \n // Apply on load\n applyDarkMode();\n \n // Re-apply on dynamic content\n var observer = new MutationObserver(function(mutations) {\n if (enabled && !document.getElementById('universal-dark-mode-style')) {\n applyDarkMode();\n }\n });\n observer.observe(document.head, { childList: true });\n \n console.log('[Universal Dark Mode] Loaded - Press Alt+Shift+D to toggle');\n})();", "Universal Dark Mode"); } } catch(__e) { console.warn('[Userscript:Universal Dark Mode]', __e); } })(); })();
Skip to content

Repository files navigation

Deep Research Agent

A plugin-based agentic research framework built on LangGraph. It decomposes a research question into sub-questions, retrieves and scrapes information from multiple sources, checks its own work for gaps, and synthesizes a structured, cited report — with every search provider, scraper, and LLM backend swappable through configuration.

PythonLangGraphStatus

Table of Contents

Why This Exists

Most "research agent" scripts hard-code a single search API and a single model. This framework treats every external dependency — search, scraping, and LLM inference — as a pluggable, prioritized, fallback-capable tool, so a provider outage or missing API key degrades gracefully instead of breaking the run. The orchestration itself (planning → retrieval → reflection → synthesis) is a LangGraph state machine that knows nothing about which concrete tools it's calling.

Features

  • Plugin architecture — swap search providers, scrapers, and LLM backends entirely through YAML configuration, no code changes required.
  • Automatic fallback — each tool category (search, scraper, LLM) has a priority-ordered fallback chain; if the top tool fails, the next one is tried automatically.
  • Iterative research loop — the agent plans, retrieves, reflects on what's missing, and loops back to retrieval until the report is complete or max_loops is hit.
  • Structured citations — reports carry inline [n] citations backed by a Citation model with URL, title, excerpt, and access timestamp.
  • Model routing — routes planning/reflection/synthesis to different LLM tiers (fast / balanced / powerful) via LiteLLM, so cheaper models handle cheaper steps.
  • Drop-in custom tools — add a file to tools/custom/, subclass a base tool, and it's auto-discovered on startup.

Architecture

The framework is organized into four layers:

  1. Core orchestration (core/) — a tool-agnostic LangGraph state machine that drives the research loop.
  2. Tool registry (registry/) — discovers, prioritizes, and selects tools, with fallback-chain resolution.
  3. Tool implementations (tools/) — concrete search, scraper, LLM, and custom tool adapters.
  4. Data models (models/) — Pydantic schemas for citations and tool I/O, so every tool boundary is type-checked.

Research Workflow

Query → Planning → Retrieval → Reflection → Synthesis → Report
↑ ↓
└─ (if gaps) ─┘
  1. Planning — decomposes the query into 3–5 sub-questions using a fast LLM.
  2. Retrieval — searches and scrapes content for each sub-question, trying tools down the fallback chain as needed.
  3. Reflection — evaluates completeness against the original query and identifies remaining gaps.
  4. Synthesis — once the reflection step is satisfied (or max_loops is reached), a powerful LLM generates the final cited report.

Installation

Prerequisites

  • Python 3.8+
  • pip (or conda)

Setup

# 1. Clone the repository
git clone https://github.com/QuantumWars/DeepResearchAgent.git
cd DeepResearchAgent
# 2. Install dependencies
pip install -r requirements.txt
# 3. Install Playwright browsers (needed for JS-heavy sites)
playwright install
# 4. Configure API keys
cp .env.example .env

Then edit .env with the providers you plan to use:

# Search (at least one required)
TAVILY_API_KEY=your_tavily_key_here
SERPER_API_KEY=your_serper_key_here
# LLM (at least one required)
OPENAI_API_KEY=your_openai_key_here
# or
ANTHROPIC_API_KEY=your_anthropic_key_here

Quick Start

Command Line

# Basic research query
python main.py "What is quantum computing?"# Save the report to a file
python main.py "Explain climate change" --output report.md
# Cap the number of retrieval/reflection loops
python main.py "AI safety research" --max-loops 2
# Verbose logging for debugging
python main.py "Machine learning basics" --verbose

Programmatic Usage

fromcore.orchestratorimportResearchOrchestratororchestrator=ResearchOrchestrator("config/tool_config.yaml")
result=orchestrator.research(
query="What is quantum computing?",
max_loops=3
)
print(result.report)
print(f"Sources: {len(result.sources)}")
print(f"Tool calls: {len(result.execution_log)}")
result.save("reports/quantum_computing.md")
forcitationinresult.get_citations():
print(f"[{citation.id}] {citation.title} - {citation.url}")

Using Custom Tools

fromcore.orchestratorimportResearchOrchestratorfromregistry.base_toolimportBaseCustomToolclassPDFExtractor(BaseCustomTool):
name="pdf_extractor"description="Extracts text from PDF files"defexecute(self, input_data):
url=input_data.get("url")
# ... extraction logic ...return {"success": True, "content": extracted_text}
orchestrator=ResearchOrchestrator()
result=orchestrator.research(
query="Research question",
custom_tools=[PDFExtractor()]
)

Configuration Guide

All tool selection, priority, and fallback behavior lives in config/tool_config.yaml.

Search Tools

search_tools:
fallback_chain: # tried in order until one succeeds
- tavily
- serpertavily:
enabled: truepriority: 10# higher = tried firstapi_key: env:TAVILY_API_KEYextra_params:
search_depth: basic # or 'advanced'max_results: 5

Available: tavily (fast, AI-optimized — recommended), serper (Google Search API).

Scraper Tools

scraper_tools:
fallback_chain:
- trafilatura
- playwrighttrafilatura:
enabled: truepriority: 10extra_params:
include_tables: truededuplicate: trueplaywright:
enabled: truepriority: 5extra_params:
headless: truetimeout: 30000wait_for: networkidle

Available: trafilatura (fast, lightweight, for standard sites), playwright (full browser automation, for JS-heavy sites).

LLM Tools

llm_tools:
routing:
fast: gpt-3.5-turbo # planningbalanced: gpt-4-turbo-preview # reflectionpowerful: gpt-4 # synthesisprovider: litellmextra_params:
temperature: 0.7max_tokens: 4000

Supported providers (via LiteLLM): OpenAI, Anthropic, Cohere, and anything else LiteLLM speaks.

Workflow Settings

workflow:
max_loops: 3# maximum research iterationsmax_documents: 20# maximum documents to retrievemax_scrape_size: 102400# max bytes per page (100KB)tool_timeout: 30# timeout per tool call (seconds)

Environment Variables

config/tool_config.yaml references secrets with an env:VAR_NAME syntax, resolved from .env:

TAVILY_API_KEY=tvly-xxxxx
SERPER_API_KEY=xxxxx
OPENAI_API_KEY=sk-xxxxx
ANTHROPIC_API_KEY=sk-ant-xxxxx
COHERE_API_KEY=xxxxx

Adding Custom Tools

Drop a file in tools/custom/ and subclass the appropriate base:

# tools/custom/my_tool.pyfromregistry.base_toolimportBaseCustomToolimportlogginglogger=logging.getLogger(__name__)
classMyCustomTool(BaseCustomTool):
"""Description of what your tool does."""name="my_tool"description="Custom tool for specific task"def__init__(self, api_key=None, extra_params=None):
self.api_key=api_keyself.config=extra_paramsor {}
logger.info(f"Initialized {self.name}")
defexecute(self, input_data):
try:
result=self._process(input_data)
return {"success": True, "data": result}
exceptExceptionase:
logger.error(f"{self.name} failed: {e}")
return {"success": False, "error": str(e)}

No manual registration needed — the registry discovers and registers tools on startup. Optionally add config for it under custom_tools: in tool_config.yaml. Pick the right base class:

  • BaseSearchTool — search providers
  • BaseScraperTool — web scrapers
  • BaseLLMTool — LLM integrations
  • BaseCustomTool — everything else

See examples/custom_tool_example.py for a complete walkthrough.

API Reference

ResearchOrchestrator

classResearchOrchestrator:
def__init__(self, config_path: str="config/tool_config.yaml")
defresearch(
self,
query: str,
custom_tools: Optional[List[BaseTool]] =None,
max_loops: int=3
) ->ResearchResult

ResearchResult

classResearchResult:
report: str# markdown report with citationssources: List[dict] # source documentsexecution_log: List[dict] # tool execution historydefsave(self, filepath: str) ->Nonedefget_citations(self) ->List[Citation]

ToolRegistry

classToolRegistry:
defregister_tool(self, tool_instance: BaseTool, category: str, priority: Optional[int] =None) ->Nonedefget_tool(self, category: str, name: Optional[str] =None) ->Optional[BaseTool]
defget_tool_chain(self, category: str) ->List[BaseTool]
@classmethoddeffrom_config(cls, config_path: str) ->"ToolRegistry"defdiscover_tools(self, tools_directory: str="tools") ->int

Project Structure

DeepResearchAgent/
├── core/ # Core orchestration layer
│ ├── orchestrator.py # ResearchOrchestrator / ResearchResult
│ ├── graph.py # LangGraph workflow definition
│ ├── workflow_nodes.py # Planning/retrieval/reflection/synthesis nodes
│ └── state.py # Shared research state
├── registry/ # Tool registry system
│ ├── tool_registry.py # Discovery, priority, fallback chains
│ └── base_tool.py # Abstract base classes
├── tools/ # Tool implementations
│ ├── search/ # tavily_search.py, serper_search.py
│ ├── scraper/ # trafilatura_scraper.py, playwright_scraper.py
│ ├── llm/ # litellm_tool.py
│ └── custom/ # Your custom tools go here
├── models/ # Pydantic schemas
│ └── tool_schemas.py
├── config/
│ └── tool_config.yaml
├── utils/ # Config loading, logging, formatting helpers
├── examples/ # Runnable usage examples
├── main.py # CLI entry point
├── requirements.txt
└── .env.example

Testing

# End-to-end execution smoke test
python test_execution.py
# Usage examples double as integration checks
python examples/basic_research.py
python examples/custom_tool_example.py
python examples/test_orchestrator.py

See TEST_RESULTS.md and test_report.md for the latest recorded run output.

Troubleshooting

Tool 'tavily' failed: API key not found Check that .env exists, contains TAVILY_API_KEY=tvly-xxxxx, and that the process was restarted after adding it.

All search tools failing Verify API key validity, internet connectivity, and rate limits; rerun with --verbose to see per-tool logs.

Playwright scraper timing out Raise timeout under playwright.extra_params in tool_config.yaml, fall back to trafilatura for simpler sites, or check whether the target blocks automation.

ModuleNotFoundError: No module named 'langgraph'pip install -r requirements.txt.

playwright._impl._api_types.Error: Executable doesn't existplaywright install.

Empty report Make the query more specific, confirm at least one search tool is healthy, check the verbose logs, or raise --max-loops.

Research is slow Lower max_loops, reduce max_results on search tools, use faster models in llm_tools.routing, or trim the fallback chains.

Enable verbose logging for any of the above:

python main.py "query" --verbose

Contributing

Contributions are welcome. Areas that could use help:

  • Additional search providers (Brave, Bing, etc.)
  • More scraper implementations
  • New custom node implementations
  • Performance optimizations
  • Documentation improvements

Design docs live under .kiro/specs/deep-research-framework/ if you want the full requirements/design/task breakdown behind the current implementation.

License

No formal license has been selected yet. All rights reserved by QuantumWars until one is chosen — open an issue if you need clarification for a specific use case.

About

Plugin-based LangGraph research agent that plans, retrieves, reflects, and synthesizes cited reports — with swappable search, scraper, and LLM providers and automatic fallback.

Resources

Stars

0 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages