Hari Nagarajan — Security Engineer, Marine veteran, and recovering physicist.
I lead threat analysis, incident response, and AI-driven response automation at Amazon Leo Physical & IP Security. Outside of work, I give LLMs offensive security tools without guardrails and see what happens.
Two open-source projects worth pointing at — both Rust, both terminal-native.
The thesis is simple: a chat window is the wrong shape for security work. You want a real terminal — split panes, log scrollback, copy-paste, the whole thing — with an agent embedded in it as a co-pilot, not a wrapper around it.
seval gives the model real shell, fs and http tools and a session log it can write to. You point it at a scope, narrate the goal, and it works until it gets stuck or finishes. Bring your own LLM (OpenAI, Anthropic, local Ollama).
Most “AI security tools” are wrappers around chat completions that can’t actually do anything. seval gives the model real access — a shell, a filesystem, HTTP — and lets it work the way a human operator would, just faster and without getting bored at 2am.
eugene treats a Pi like a tiny operator you can leave behind. Boot it, hand it a scope, and it cycles through host-discovery, service fingerprinting, low-hanging credentials, persisting everything to a local SQLite so you can resume after a reboot or replay the run.
The agent loop is rig-core; the tools are whatever Kali already ships. The interesting bit is the human-in-the-loop primitive — eugene pauses at every meaningful finding and waits for a one-word ack before continuing. So it’s autonomous, but never quietly autonomous.
Eugene uses a planner/executor split. The orchestrator breaks work into parallel tasks, and specialist executor agents run each one independently via tokio + semaphore-bounded concurrency. Everything is tracked in SQLite with FTS5 full-text search.
You can control eugene from Telegram, the CLI, or let it run on a cron schedule. Every run, task, and finding is persisted locally. The TUI dashboard gives you a full-screen ratatui view with progress gauge, findings table, and activity log.