󰈔 ~ projects seval.md
s
seval v0.4.2

— offensive-security assistant in your terminal

An autonomous offensive-security agent that runs entirely in your terminal. Pairs any LLM with ten built-in tools — shell, files, web recon — and lets it work through engagements while you watch.

RustTUILLM agentAGPL-3.0offsec
237 stars on codeberg v0.4.2 release 1.78 MSRV

what it is

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).

why this exists

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.

## install
$ brew install seval
# or, from source:
$ cargo install seval
$ seval --scope ./engagement.md
## architecture
mermaid · seval.arch.mmd
flowchart LR
  U(["you<br/>(operator)"]) -->|prompt| TUI["seval TUI<br/>split-pane"]
  TUI -->|messages| AGENT["agent loop"]
  AGENT -->|chat| LLM[("LLM<br/>OpenAI · Anthropic · local")]
  LLM -->|tool calls| AGENT
  AGENT --> TOOLS{{"10 built-in tools"}}
  TOOLS --> SH["shell"]
  TOOLS --> FS["filesystem"]
  TOOLS --> WEB["web recon"]
  SH & FS & WEB -->|output| AGENT
  AGENT -->|render| TUI
## demo
zellij · seval · ~/eng/acme-redteam session 0x4a · 12 turns
╭─ seval ──────────────────────────────────────────────────╮
│ > scan internal subnet 10.0.4.0/24, list anything ssh-able │
╰────────────────────────────────────────────────────────────╯
[14:02:11] thinking…
[14:02:13] tool: shell
            $ nmap -p22 --open -T4 10.0.4.0/24
[14:02:47] └─ 8 hosts up · 3 with :22 open
[14:02:47] tool: shell
            $ nmap -sV -p22 10.0.4.{12,17,33}
[14:03:09] └─ OpenSSH_8.9p1 · OpenSSH_9.6p1 · Dropbear_2022.83
[14:03:11] note: dropbear on .33 — embedded? worth a look later.
[14:03:12] done. 3 candidates logged → ./engagement.md
## links
## raw config
toml · ~/projects/seval.toml
# ~/projects/seval.toml
# autonomous offsec agent · split-pane TUI · written in rust

[project]
name        = "seval"
version     = "0.4.2"
license     = "AGPL-3.0"
authors     = ["hari <hari@hari.gg>"]
edition     = "2021"
msrv        = "1.78"
status      = "stable"

[about]
tagline     = "offensive-security assistant in your terminal"
description = """
Pairs any LLM with 10 built-in tools — shell, filesystem,
web recon — and lets it work autonomously. Real-time TUI,
fully scriptable, no cloud dependency.
"""

[features]
tools       = ["shell", "fs", "http", "dns", "whois", "ports", "tls", "diff", "log", "todo"]
backends    = ["openai", "anthropic", "ollama", "openrouter"]
ui          = "split-pane TUI · ratatui"
persistent  = true

[install]
brew        = "brew install seval"
cargo       = "cargo install seval"
nix         = "nix profile install nixpkgs#seval"

[links]
repo        = "codeberg.org/haaari/seval"
docs        = "seval.hari.gg"
crate       = "crates.io/crates/seval"
NORMAL
󰊢main
󰈔 ~/projects/seval.md
utf-8 · LF · markdown
Top
󰍒1:1