Add DeepReadFile tool for multi-format document support

[ssgamingop]

Description

This PR introduces the DeepReadFile tool to the shared_tools module, enabling agents to read and process non-plain-text documents. While the existing ReadFile tool is excellent for source code and text files, agents were previously "blind" to common business formats like PDFs and Word documents.

This implementation leverages the markitdown library to provide high-quality Markdown extractions from various formats.

Changes

• Created shared_tools/DeepReadFile.py using markitdown.
• Exported DeepReadFile in shared_tools/__init__.py.
• Integrated the tool into the Virtual Assistant (General Agent) and Deep Research Agent.
• Supports: PDF, DOCX, XLSX, PPTX, HTML, and more.

Why this is needed

Users frequently upload:

• Research papers (PDF)
• Project specs (DOCX)
• Data exports (XLSX)

This tool allows the swarm to ingest this data directly into their context, significantly expanding the utility of the Deep Research and Virtual Assistant agents.

How to test

1. Upload a PDF or DOCX to the workspace.
2. Ask the General Agent: "Read the content of [path/to/file.pdf] and summarize it."
3. Verify the agent successfully extracts and summarizes the text.

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 440d548e8d

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".