Skip to main content

Phase 1: Research with /mem8:research

Deep codebase analysis using parallel sub-agents to understand architecture and patterns.

How It Works​

What You Get​

Structured Research Document:

# Research: How Does Authentication Work?

## Summary
- Token-based JWT authentication
- Middleware validates on every request
- Redis stores active sessions

## Detailed Findings

### Authentication Flow (auth/middleware.py:45-67)
The middleware intercepts requests and validates JWT tokens...

[file references with line numbers]

## Architecture Insights
- Stateless JWT design allows horizontal scaling
- Refresh token rotation prevents token theft
- Rate limiting per user prevents abuse

## Code References
- `auth/middleware.py:45` - JWT validation
- `auth/tokens.py:23` - Token generation
- `auth/redis_store.py:12` - Session storage

Key Features​

  • Parallel sub-agents explore different aspects simultaneously
  • Full file context - reads mentioned files completely before spawning agents
  • Persistent memory - research saved to memory/shared/research/
  • Git metadata - captures commit, branch, researcher for context
  • File references - clickable links to specific lines of code