VRSEN · ssgamingop · May 6, 2026 · May 6, 2026 · May 6, 2026 · May 6, 2026
diff --git a/deep_research/deep_research.py b/deep_research/deep_research.py
@@ -2,6 +2,7 @@
 from agency_swarm.tools import WebSearchTool, IPythonInterpreter
 from openai.types.shared import Reasoning
 from virtual_assistant.tools.ScholarSearch import ScholarSearch
+from shared_tools import DeepReadFile
 
 from config import get_default_model, is_openai_provider
 
@@ -12,7 +13,7 @@ def create_deep_research() -> Agent:
         description="Comprehensive deep research agent that conducts thorough research on any topic.",
         instructions="./instructions.md",
         files_folder="./files",
-        tools=[WebSearchTool(), ScholarSearch, IPythonInterpreter],
+        tools=[WebSearchTool(), ScholarSearch, IPythonInterpreter, DeepReadFile],
         model=get_default_model(),
         model_settings=ModelSettings(
             reasoning=Reasoning(effort="high", summary="auto") if is_openai_provider() else None,

diff --git a/docs_agent/tools/CreateDocument.py b/docs_agent/tools/CreateDocument.py
@@ -35,7 +35,8 @@ class CreateDocument(BaseTool):
     - .source.html file (the canonical source of truth)
 
     Markdown workflow creates:
-    - .md file only (no .docx or .pdf generation)
+    - .md file
+    - .source.html file (generated from Markdown to enable PDF/DOCX conversion)
 
     HTML is used as the source format because it provides:
     - Full styling control (fonts, colors, spacing, etc.)
@@ -141,10 +142,25 @@ def run(self):
 
     def _create_markdown(self, doc_name, project_dir, markdown_value):
         md_path = project_dir / f"{doc_name}.md"
-        if md_path.exists() and not self.overwrite:
-            return f"Error: Document '{doc_name}' already exists in project '{self.project_name}'. Use overwrite=True to replace it, or choose a different document name."
+        source_path = project_dir / f"{doc_name}.source.html"
+
+        # Guard: Respect overwrite flag for BOTH files
+        if not self.overwrite:
+            if md_path.exists():
+                return f"Error: Markdown file '{md_path.name}' already exists in project '{self.project_name}'. Use overwrite=True to replace it."
+            if source_path.exists():
+                return f"Error: HTML source '{source_path.name}' already exists in project '{self.project_name}'. Creating this Markdown document would overwrite it. Use overwrite=True to proceed."
 
         md_path.write_text(markdown_value, encoding="utf-8")
+
+        # Also create a .source.html version from Markdown to enable PDF/DOCX conversion
+        from markdown_it import MarkdownIt
+        md_parser = MarkdownIt()
+        html_body = md_parser.render(markdown_value)
+        html_full = f"<!DOCTYPE html><html><head><meta charset=\"UTF-8\"></head><body>{html_body}</body></html>"
+
+        source_path.write_text(_ensure_ua_reset(html_full), encoding='utf-8')
+
         if not md_path.exists():
             return f"Error: Markdown generation failed for document '{doc_name}'."
         md_size = md_path.stat().st_size
@@ -157,10 +173,11 @@ def _create_markdown(self, doc_name, project_dir, markdown_value):
 
 Files created:
   - {md_path.name} ({md_size:,} bytes) [Markdown source]
+  - {source_path.name} [HTML source for conversion]
 
 Path: {md_path}
 
-Note: Markdown workflow only creates a .md file and does not generate .docx or .pdf files."""
+Note: You can now convert this document to PDF or DOCX using the ConvertDocument tool."""
 
 def _build_html_preview_image(html_content: str, base_dir: Path):
     """Render a preview JPEG of the HTML document.

diff --git a/pyproject.toml b/pyproject.toml
@@ -47,6 +47,7 @@ dependencies = [
     "cairosvg",
     "weasyprint",
     "html2text",
+    "markdown-it-py",
     "playwright",
     # AI / media generation
     "google-genai",

diff --git a/requirements.txt b/requirements.txt
@@ -42,6 +42,7 @@ beautifulsoup4
 cairosvg
 weasyprint
 html2text
+markdown-it-py
 google-genai
 opencv-python-headless
 moviepy<2

diff --git a/shared_tools/DeepReadFile.py b/shared_tools/DeepReadFile.py
@@ -0,0 +1,55 @@
+import os
+from typing import Optional
+from agency_swarm.tools import BaseTool
+from pydantic import Field
+from markitdown import MarkItDown
+
+class DeepReadFile(BaseTool):
+    """
+    A tool to read and convert various file formats into Markdown.
+    Supports PDF, DOCX, XLSX, PPTX, HTML, and common image formats (via OCR if configured).
+
+    Use this tool when you need to understand the content of a non-plain-text file.
+    """
+
+    file_path: str = Field(..., description="The absolute path to the file to read and convert.")
+
+    def run(self):
+        try:
+            # Ensure the path is absolute
+            abs_path = os.path.abspath(self.file_path)
+
+            if not os.path.exists(abs_path):
+                return f"Error: File does not exist at {abs_path}"
+
+            if not os.path.isfile(abs_path):
+                return f"Error: {abs_path} is not a file."
+
+            # Initialize MarkItDown
+            md = MarkItDown()
+
+            # Convert the file
+            result = md.convert(abs_path)
+
+            if not result or not result.text_content:
+                return f"Warning: Conversion successful but no text content was extracted from {os.path.basename(abs_path)}."
+
+            content = result.text_content
+
+            # Add a header to indicate the source
+            header = f"--- Content of {os.path.basename(abs_path)} ---\n\n"
+            return header + content
+
+        except Exception as e:
+            return f"Error converting file {self.file_path}: {str(e)}"
+
+if __name__ == "__main__":
+    # Test with a simple text file first
+    test_file = "test_deep_read.txt"
+    with open(test_file, "w") as f:
+        f.write("# Test Heading\nThis is a test file for DeepReadFile.")
+
+    tool = DeepReadFile(file_path=os.path.abspath(test_file))
+    print(tool.run())
+
+    os.remove(test_file)
diff --git a/shared_tools/__init__.py b/shared_tools/__init__.py
@@ -1,7 +1,15 @@
 from shared_tools.CopyFile import CopyFile
+from shared_tools.DeepReadFile import DeepReadFile
 from shared_tools.ExecuteTool import ExecuteTool
 from shared_tools.FindTools import FindTools
 from shared_tools.ManageConnections import ManageConnections
 from shared_tools.SearchTools import SearchTools
 
-__all__ = ["CopyFile", "ExecuteTool", "FindTools", "ManageConnections", "SearchTools"]
+__all__ = [
+    "CopyFile",
+    "DeepReadFile",
+    "ExecuteTool",
+    "FindTools",
+    "ManageConnections",
+    "SearchTools",
+]
diff --git a/virtual_assistant/virtual_assistant.py b/virtual_assistant/virtual_assistant.py
@@ -8,7 +8,14 @@
 from dotenv import load_dotenv
 
 from config import get_default_model, is_openai_provider
-from shared_tools import CopyFile, ExecuteTool, FindTools, ManageConnections, SearchTools
+from shared_tools import (
+    CopyFile,
+    DeepReadFile,
+    ExecuteTool,
+    FindTools,
+    ManageConnections,
+    SearchTools,
+)
 
 load_dotenv()
 
@@ -33,6 +40,7 @@ def create_virtual_assistant() -> Agent:
             PersistentShellTool,
             IPythonInterpreter,
             CopyFile,
+            DeepReadFile,
             ExecuteTool,
             FindTools,
             ManageConnections,