Index Project Command

Index the current project directory using Infinite Memory's semantic search capabilities.

Your Task

1. Get the current working directory:

   pwd
   

2. Call the index_project MCP tool with the absolute path:

   • Use the path from step 1
   • The tool will:
     • Scan all supported file types (.py, .js, .ts, .md, .txt, .json, .yaml, .yml)
     • Generate embeddings using sentence-transformers
     • Store in local Pixeltable database
     • Respect .gitignore patterns
     • Use hash-based change detection (only re-index changed files)

3. Show progress and results to the user:

   • Inform user indexing started
   • When complete, show:
     • Files processed
     • Files indexed/inserted
     • Files updated
     • Files skipped (no changes)
     • Files failed (if any)
     • Duration

4. Suggest next steps:

   • "Your code is now searchable!"
   • Suggest trying: "How does [key feature] work?" or "/search [query]"

Example Output

Indexing /Users/sarah/projects/api-service...

✅ Complete!
- 487 files processed
- 482 files indexed
- 3 files skipped (no changes)
- 2 files failed (permission denied)
- Completed in 0.31s
- Storage: ~2.1 MB

Your code is now searchable! Try:
- "How does authentication work?"
- "/search database connection pooling"

When to Use

• First time using Infinite Memory in a project
• After major codebase refactoring
• When you want to ensure all files are indexed
• After pulling significant changes from version control

Notes

• The MCP tool automatically skips:
  • Binary files
  • System directories (/etc, /sys, /proc, /dev)
  • Files excluded by .gitignore
• Indexing is incremental - unchanged files are skipped automatically
• Large projects (1000+ files) typically index in <1 second