437 Assistant Upload 24cbbd6d

Before proceeding, ALWAYS INVOKE the pinecone:assistant skill before commencing workflow. This will allow correct resolution of plugin root directory for scripts being run.

Script paths are relative to the plugin root directory. Run scripts with: uv run skills/assistant/scripts/script_name.py [arguments]

Overview

Upload files or entire repository contents to a Pinecone Assistant. This command helps users provide knowledge to their assistants by uploading documentation and text-based files.

Key Principle: Users tell Claude what repository or directory to upload, and Claude scans and uploads the relevant files.

Important: Pinecone Assistant works with documentation and data files only, not source code. Focus on uploading:

• README files
• Documentation directories (markdown, PDF, Word)
• User guides and API docs
• Design documents and specifications
• Meeting notes and reports
• JSON data files

Supported: .docx, .json, .md, .pdf, .txt

Not supported: Source code files (.py, .js, .ts, .java, etc.) as Assistant is optimized for natural language documents, not code syntax.

Workflow

1. Parse Arguments:

   • assistant: Name of the assistant to upload to (required)
   • source: File path, directory path, or repository to upload (required)
   • patterns: Comma-separated glob patterns (optional, defaults to common doc/code files)
   • exclude: Directories to exclude (optional, defaults to node_modules, .venv, etc.)
   • metadata: Additional JSON metadata (optional)

2. If Arguments Missing:

   • If no assistant: List assistants and use AskUserQuestion to let user select
   • If no source: Ask user for the repository or directory path
     • Suggest current directory (.)
     • Suggest common paths like ./docs, ./src, etc.

3. Understand Source Type:

   • Single file: Upload it directly
   • Directory/Repository: Scan for relevant files
   • Current repo (. or ./): Use current working directory

4. Ask User About File Selection:

   • Use Glob to preview what files will be uploaded
   • Show user the count and types of files found
   • Use AskUserQuestion to confirm or let them customize patterns:
     • "Upload all documentation files (recommended)" - *.md, *.txt, *.pdf
     • "Only upload markdown files" - *.md
     • "Only upload PDFs" - *.pdf
     • "Custom documentation pattern"
   • If code files detected in directory: Actively warn user and suggest filtering them out

5. Execute Upload Script:

   uv run skills/assistant/scripts/upload.py \
     --assistant "assistant-name" \
     --source "/path/to/repo" \
     --patterns "*.md,*.py,*.js" \
     --exclude "node_modules,.venv,.git"
   

6. Display Progress and Results:

   • The script shows a progress bar during upload
   • Summary table with uploaded/failed counts
   • Next steps suggestions

7. Post-Upload Actions:

   • Remind user that files are being processed
   • Suggest trying a chat query after a moment
   • Offer to run /pinecone:assistant-chat now

Default File Patterns

The upload script includes only documentation and data files by default:

• **/*.md - Markdown files
• **/*.txt - Text files
• **/*.pdf - PDF documents
• **/*.docx - Word documents
• **/*.json - JSON data files

Code files are NOT supported: Do not upload .py, .js, .ts, .java, or any other source code files. Assistant is designed for natural language documents only.

Default Exclusions

These directories are excluded by default:

• node_modules
• .venv, venv
• .git
• build, dist
• __pycache__
• .next, .cache

Example Usage

Upload current repository:

/pinecone:assistant-upload assistant docs-qa source .

Upload specific directory:

/pinecone:assistant-upload assistant my-bot source ./documentation

Upload with custom patterns (documentation only):

/pinecone:assistant-upload assistant docs-bot source ./docs --patterns "*.md,*.pdf"

Upload single file:

/pinecone:assistant-upload assistant support-bot source ./README.md

Repository Upload Workflow

When user says "upload my repository" or provides a repo path:

1. Confirm the repository path with the user
2. Use Glob to preview files:

   # Example: Count files by type
   ls -R | grep -E '\.(md|py|js|ts)$' | wc -l
   

3. Show user what will be uploaded:
   • Total file count
   • Breakdown by type (e.g., "25 Markdown files, 8 PDFs, 3 text files")
   • If code files detected: Warn user that Assistant works best with documentation, suggest filtering to docs only
4. Ask for confirmation with AskUserQuestion
5. Execute the upload script
6. Show progress and results

Warning for Code Files: If user's directory contains .py, .js, .ts, .java or other code files, actively warn and filter them out:

⚠️  Found 50 Python files and 30 JavaScript files in this directory.

Pinecone Assistant does not support source code - it's designed for documentation only.

I found these documentation files instead:
- 15 Markdown files (READMEs, docs)
- 5 PDF files
- 3 text files

I'll upload only the documentation files. Continue?

Do not give users the option to upload code files - always filter them out automatically.

Metadata Best Practices

Encourage users to add metadata for better organization:

--metadata '{"source":"github","repo":"owner/repo","branch":"main"}'

Benefits:

• Track where files came from
• Filter by source later
• Debug and maintain knowledge base

Interactive Mode Example

User: /pinecone:assistant-upload

Claude:

1. Lists assistants, asks user to select
2. Asks: "What repository or directory should I upload?"
3. User provides path: /Users/arjun/my-project
4. Claude scans and reports: "Found 45 documentation files: 25 Markdown, 15 PDFs, 5 text files"
5. If code files found: "⚠️ Also found 50 Python files. Assistant works with documents only - I'll skip the code files."
6. Claude asks: "Should I upload all documentation files?" (uses AskUserQuestion)
7. User confirms
8. Claude executes upload script with --patterns "*.md,*.txt,*.pdf"
9. Claude shows progress and results

Script Location

skills/assistant/scripts/upload.py

Troubleshooting

Error: PINECONE_API_KEY not set

• Remind user to export PINECONE_API_KEY before starting Claude Code

Error: Assistant not found

• List available assistants
• Check for typos in assistant name

Error: Path does not exist

• Verify the path with user
• Check current working directory
• Use absolute paths if relative paths fail

No files found

• Check if patterns match file types in directory
• Verify directory isn't empty
• Check exclusion patterns aren't too broad

Upload failures

• Check file formats are supported (PDF, TXT, MD, etc.)
• Verify files aren't corrupted
• Check network connectivity
• Try uploading in smaller batches

Large repository warnings

• If >100 documentation files, ask user if they want to be more selective
• Suggest uploading specific documentation directories (e.g., ./docs only)
• Offer to exclude examples or auto-generated docs

Code files detected

• Always filter out code files automatically
• Warn user that code files were found and excluded
• Explain that Assistant is for documents only

Post-Upload Guidance

After successful upload:

1. Remind about processing time: "Files are being indexed and will be available shortly"
2. Suggest a test query: Offer to run /pinecone:assistant-chat with a sample question
3. Monitor status: Suggest checking assistant status with list command

Related Commands

• /pinecone:assistant-chat - Chat with the assistant using uploaded knowledge
• /pinecone:assistant-context - Search uploaded files
• List assistants: Run uv run skills/assistant/scripts/list.py