How harshithgraj/red-team-ai-agent works
Red Team AI Agent is a local proposal stress-testing CLI. Users supply a proposal by file, inline text, or interactive input, choose a persona and standard or debate mode, and receive AI-generated critique; reports and debate logs can be saved as Markdown. The principal orchestration and Groq request logic are in redteam.py. The README also documents an HTML findings portfolio, but its data-loading or runtime links are not established by the sampled source, so those are left unwired.
Main entry points: redteam.py Standard Analysis
Components 10
Proposal Input
- CLI Entry
redteam.py - Proposal Text
redteam.py
Critique Workflows
- Persona Prompts
redteam.py - Standard Analysis
redteam.py - Debate Mode
redteam.py
AI Integration
- Groq Chat API
Review Outputs
- Critique Report
redteam.py - Debate Log
redteam.py - Findings Portfolio
redteam-portfolio.html
Other components
- Proponent
Connections 12
- Proponent → CLI Entry: supplies proposal
- CLI Entry → Proposal Text: reads text
- CLI Entry → Standard Analysis: starts analysis
- CLI Entry → Debate Mode: starts debate
- Standard Analysis → Persona Prompts: selects prompt
- Debate Mode → Persona Prompts: selects prompt
- Standard Analysis → Groq Chat API: sends messages
- Debate Mode → Groq Chat API: sends history
- Groq Chat API → Standard Analysis: returns critique
- Groq Chat API → Debate Mode: returns response
- Standard Analysis → Critique Report: saves report
- Debate Mode → Debate Log: saves transcript