Auggie Index Command

Index the current codebase to enable context-aware search and assistance.

What This Command Does

1. Scans the current project directory for code files
2. Uploads file contents to the Auggie Context Engine for indexing
3. Saves the index state to auggie-context.json for persistence between sessions
4. On subsequent runs, performs incremental updates (only changed files)

Usage

Run this command:

• After initial setup (/auggie-setup)
• After making significant changes to your codebase
• If search results seem stale or incomplete

Execution Steps

Step 1: Check authentication

First verify Auggie credentials are configured:

auggie token print

If not authenticated, direct user to run /auggie-setup first.

Step 2: Index the codebase

Explain to the user that you will use the Auggie Context Engine SDK to:

1. Check for existing index: Look for auggie-context.json in the project root
2. Load or create index:
   • If auggie-context.json exists: Load it with DirectContext.importFromFile()
   • If not: Create new with DirectContext.create()
3. Gather files: Collect all code files (respecting .gitignore patterns)
4. Index files: Call addToIndex() with file contents
5. Save state: Call exportToFile('auggie-context.json')

Step 3: Confirm completion

Tell the user:

• How many files were indexed
• That the index is saved to auggie-context.json
• They can now use /auggie for context-aware assistance

Technical Details

The index uses Auggie's DirectContext API:

• Files are uploaded and processed by Auggie's context engine
• Index state (file checksums, blob references) is stored in auggie-context.json
• Incremental indexing: Only new/changed files are re-uploaded
• The index file should be added to .gitignore (contains project-specific data)

Files Indexed

By default, indexes common code file types:

• TypeScript/JavaScript: .ts, .tsx, .js, .jsx
• Python: .py
• Ruby: .rb
• Go: .go
• Rust: .rs
• And more...

Files in .gitignore and common exclusions (node_modules/, .git/, etc.) are skipped.

Notes

• First-time indexing may take a few minutes for large codebases
• Re-running after changes is fast (incremental updates)
• The auggie-context.json file is project-specific and should not be committed