Red Teamer Seeks Advice on Local LLM Harness for Authorized Security Exercises
CivilBug4007 · reddit · 2026-09-04
A cybersecurity practitioner working on BFSI projects is asking Reddit for help building an AI-powered red teaming harness that orchestrates recon, enumeration, vulnerability analysis, tool execution, evidence collection, and reporting with human oversight at key stages. Because project data can't leave the premises, everything must run locally. He asks which local LLMs are best for tool calling, coding, and reasoning (mentioning Orion 1.5 and Qwen-class models), what Mac hardware to target (RAM for 30B+ models, Mac Studio vs MacBook Pro), and which frameworks people use for autonomous security workflows. Priority: privacy, reliable tool use, and reasoning over benchmark scores.
More from coding & agent
- Dev: Codex optimized for hours against the wrong label column, 'deeply uncurious' — ivan_bezdomny · 2026-09-04
- No-code dev shares full AI pipeline for cozy game: Opus, Gemini, Meshy, Claude, Unity — Gambo7592 · 2026-09-04
- 43 Projects From Singapore's ChatGPT Sites Hackathon Open for Voting — gabrielchua · 2026-09-04
- img2threejs explained: turning one photo into editable Three.js code with 80k-180k tokens — maier_ak · 2026-09-04
- Dev Adds Voice Chat to 'Ghost' AI Assistant So It Can Check His Agent Orchestrator — BLUECOW009 · 2026-09-04
- Practitioner: Tool-Call Trajectories Are Higher-Signal Than CoT for Agent Monitoring — joshua_saxe · 2026-09-04