ARTEX vs AgenticPentest:a gated platform against a scoped red team workbench.
AgenticPentest (AaronCaiii/AgenticPentest) is an MIT-licensed red team testing workbench written mostly in JavaScript. It is an independent project, not a fork of ARTEX, and it does not claim to come from ARTEX. ARTEX is an AGPL-3.0 platform that lives on in forks. This page compares design, safety controls, status, published numbers and attack reporting, as of 2026-11-19.
Data as of 2026-11-19
ARTEX and AgenticPentest side by side
| Aspect | ARTEX | AgenticPentest |
|---|---|---|
| Maintainer | Autumn-27 wrote it. The original repo was deleted on 2026-10-08. RuoJi6/ARTEX is the main maintained fork | AaronCaiii, a personal GitHub account and its only listed contributor. Not a fork of ARTEX |
| Language | Go 1.26, Next.js 16 console, PostgreSQL 16 | JavaScript on Node.js with a Vue console and SQLite, plus Python setup scripts |
| License | AGPL-3.0 | MIT for its own code. Third-party tools keep their own licenses |
| Targets | Assets set per task, from domains and hosts to services and endpoints | Imported assets such as hosts, identities, data and channels, plus read-only AWS, GCP, Azure and Alibaba Cloud inventory |
| Architecture | A planner and 3 workers by default over an asset graph and an exploration graph | A Leader agent and eleven specialists, plus custom agents, on an editable workflow canvas, with STRIDE threat modeling |
| Guard and safety controls | Every tool call passes intercept rules, an LLM judge and human approval. A recording proxy keeps all traffic | Admin-set CIDR and domain scope, deny by default, and a Linux host guard for subprocess tools. State-changing actions need a one-time approval |
| Deployment | One Go binary with a web console and PostgreSQL, from release zips or Docker | Docker image for amd64 and arm64, or a source install. Web console plus an MCP server with 37 typed tools |
| Maintenance | RuoJi6/ARTEX: v0.3.30 on 2026-11-14, last push 2026-11-19, 2,290 stars (2026-11-19) | 8 releases, v1.0 on 2026-11-09 to v1.1.2 on 2026-11-19, last push 2026-11-19, 53 stars and 15 forks (2026-11-19) |
| Offsec Harness Index | Entry 01, tier Confirmed, breach stated. 14 primary reports. Key figure: 200+ organizations targeted (CrowdStrike) | Not listed. The index's 154 entries, read on 2026-11-19, have no AgenticPentest row |
| Public benchmark | Baidu Agent+ final: 1st overall (2026-09-03). Live range 67.83/100, 7th. TSecBench 89.78/100, 6th | None published. Its README has a feature table, not test scores |
An independent project
AgenticPentest's GitHub description calls it an agent-powered red team testing workbench with asset mapping, threat modeling, multi-agent workflows, cloud assessment, reports and MCP. The repo was created on 2026-11-05 under AaronCaiii, a personal account, and that account is its only listed contributor. GitHub does not mark it as a fork, its code is JavaScript and Python rather than Go, and its MIT license names AaronCaiii as the copyright holder. It is a separate tool, not a copy of ARTEX.
ARTEX does appear in the repo in two places. The README has a feature table that sets AgenticPentest next to ARTEX (RuoJi6/ARTEX), PentAGI, PentestGPT, Strix and Faraday, and it says an unconfirmed cell does not mean a missing feature. One legacy agent preset file says its API recon handoff borrows the approach of ARTEX's api-recon and ScopeSentry skills, and that it does not need ARTEX or import its scripts. That is a cited reference, not a claim of lineage.
Different designs
AgenticPentest is built as a workbench for one operator or a local team, with administrator, operator and viewer roles. Its README lists asset import and mapping, plans and tasks run by a Leader agent and eleven specialists, LLM profiles for OpenAI, Anthropic and Gemini-compatible endpoints, import of Nuclei, Nessus, Greenbone, Qualys and Rapid7 results, STRIDE threat modeling where LLM proposals need a person to accept them, and reports in six templates. It also lists modules for remote servers over SSH, listeners and sessions, and AI-drafted proof of concept code that it says still needs human review. An MCP server with 37 typed tools and scoped, expiring credentials drives the same backend as the web console.
ARTEX starts from a set of assets, keeps them in a shared graph, and lets a planner and workers explore until a goal is proven, with a person able to steer the run from its console.
Safety controls
AgenticPentest's scope guide says a team administrator sets the allowed CIDRs and domains, confirmed with a password, and that out-of-scope tasks and MCP calls are rejected. Its engagements guide adds a time window, allowed and banned techniques and stop conditions, with scope deny by default. Subprocess tools such as Nmap only run on Linux with a registered host guard, and are blocked without one. A state-changing action must pass six gates, including a one-time human approval tied to a verification card, or an auto-approve mode only inside a declared lab, and chain execution stays off unless an administrator turns it on. The app shows an authorization notice before the first account is made.
ARTEX checks each tool call with intercept rules, an LLM judge and human approval, and records all traffic through a proxy. AgenticPentest fences where and when work can run and asks a person before state-changing actions; read-only checks inside scope run without a per-call approval.
In the Offsec Harness Index
The Offsec Harness Index lists ARTEX as entry 01 with the tier Confirmed, citing 14 primary reports; its key figure, 200+ organizations targeted, comes from CrowdStrike's report of 2026-10-29.
AgenticPentest is not in the index. A read of its 154 entries on 2026-11-19 found no row for it, which is not surprising for a repo created two weeks earlier. Not being listed says nothing either way about use: the index's own note is that no attack on record is not the same as no use, and it publishes no safety rating.
Status, license and numbers
ARTEX is AGPL-3.0; AgenticPentest is MIT. AgenticPentest shipped 8 releases in its first two weeks, from v1.0 on 2026-11-09 to v1.1.2 on 2026-11-19, and its last push was on 2026-11-19. On that day it had 53 stars and 15 forks, against 2,290 stars for RuoJi6/ARTEX, which released v0.3.30 on 2026-11-14.
AgenticPentest publishes no benchmark or contest score. Its README says its demo video uses synthetic data with pre-seeded results rather than a live test, and that real cloud accounts remain unverified. ARTEX's numbers come from the Baidu Agent+ contest and TSecBench. With no number on one side, this page has nothing to set side by side and does not rank them.
Sources
- AaronCaiii/AgenticPentest on GitHub
- AgenticPentest README
- AgenticPentest releases
- AgenticPentest license
- AgenticPentest target scope guide
- AgenticPentest engagements guide
- AgenticPentest MCP guide
- AgenticPentest legacy agent presets
- Offsec Harness Index, full list
- Offsec Harness Index, ARTEX row
- Offsec Harness Index, about and method
- CrowdStrike, 2026-10-29
- RuoJi6/ARTEX on GitHub
- Baidu Agent+ and TSecBench scores, cn-sec
Related on this site
More comparisons
CompareAntiFox vs ARTEX
AntiFox is a public fork that says it was built on ARTEX and adds report delivery. What changed, how active it is and its license.
ARTEX vs PentestGPT
A Go platform with a web console against a Python command-line agent from a USENIX Security 2024 paper.
ARTEX vs Strix
A self-hosted Go platform against an Apache-2.0 Python agent with a Docker sandbox, a cloud edition and CI support.
ARTEX vs CAI
An AGPL platform still maintained in forks against CAI from Alias Robotics, now archived with a commercial successor.
ARTEX vs HexStrike AI
A self-contained agent platform against an MCP server that lets your own AI client call more than 150 security tools.
ARTEX vs PentAGI
Two self-hosted multi-agent systems written in Go: an AGPL platform kept alive in forks and VXControl's MIT system with a Docker sandbox.
ARTEX vs Cairn
Cairn is the one project the ARTEX README lists as its reference. Both grow a graph of facts and intents; one is a Go platform, the other a Python engine.
ARTEX vs CyberStrikeAI
Two Go platforms with a web console and human approval. CyberStrikeAI is Apache-2.0 and also ships WebShell management and a built-in C2.
ARTEX vs Shannon
Both AGPL-3.0. ARTEX runs open-ended assessments from a console; Shannon reads an app's source and only reports what it can prove.
ARTEX vs T3MP3ST
Both AGPL-3.0. ARTEX gates every tool call; T3MP3ST drives the coding agent you already use and has been quiet since 2026-09-08.
ARTEX vs ARIA-X
ARTEX and an independent Python rebuild: shared design ideas, different storage and control defaults, no performance ranking.
Follow the ARTEX story.
New forks, releases and reports, dated and sourced.
