Add PDF file upload support to GitHub Copilot CLI

Describe the feature or problem you'd like to solve

Add PDF file upload support to GitHub Copilot CLI

Proposed solution

GitHub Copilot CLI currently does not support uploading or analyzing PDF files through the interface, even though the underlying AI models fully support PDF analysis. Many users and workflows rely on PDF documents (technical documentation, specifications, reports, design files, etc.). This limitation creates friction and forces users to manually extract and paste content instead of directly referencing PDF files.

Benefit: Enabling PDF support would allow users to:

• Reference technical documentation and specifications without manual copying
• Analyze PDF reports and extract key information
• Include design specifications and mockups directly
• Improve workflow efficiency by reducing manual data extraction
• Align CLI capabilities with other Copilot interfaces (VS Code, GitHub.com)

Example prompts or workflows

1. copilot then @design-spec.pdf - Attach a PDF design specification and ask the agent to implement features based on it
2. copilot then @technical-docs.pdf - Reference a technical documentation PDF when writing code that needs to follow specific standards
3. copilot -p "Analyze this architecture diagram" @architecture.pdf - Use non-interactive mode to analyze PDF content directly
4. copilot then @requirements.pdf @existing-code.ts - Combine PDF requirements with code files for better context
5. copilot --resume <session-id> - Resume a session and reference related PDF files for continued work

Additional context

Why this matters:

• The underlying Claude/GPT models already support PDF analysis (multimodal capabilities)
• Users report having to manually copy-paste PDF content as a workaround
• This is a UX/interface limitation, not a model capability limitation

Similar functionality exists in:

• Other AI assistant CLIs

Expected behavior:

• Users should be able to attach PDF files using @filename.pdf syntax
• PDFs should be processed by the model for analysis and reference
• Multiple PDFs should be supportable in a single session

まず、CLI がインタラクティブなプロンプト、非インタラクティブモード -p、--resume セッションで @filename 参照をどのように処理しているかを確認します。既存のファイル添付とモデル入力のフローを追跡し、その後、単一および複数の PDF を参照して分析できること、コードファイルと組み合わせられること、セッションを再開した後に使用できることを検証します。