ARTEX vs Cairn:the reference and the platform built after it.
Cairn (oritera/Cairn) is a Python search engine for problems with a start, a goal and an unknown path, tested first on pentesting. The RuoJi6/ARTEX README names Cairn as its only reference. This page compares the two on design, license, safety controls, status, published numbers and attack reporting, as of 2026-11-17.
Data as of 2026-11-17
ARTEX and Cairn side by side
| Aspect | ARTEX | Cairn |
|---|---|---|
| Maintainer | Autumn-27 wrote it. The original repo was deleted on 2026-10-08. RuoJi6/ARTEX is the main maintained fork | oritera, a GitHub user account |
| Started | First commit on 2026-07-26 | Repo created 2026-04-19 |
| Language | Go 1.26, Next.js 16 console, PostgreSQL 16 | Python 3.12 or newer |
| License | AGPL-3.0 | AGPL-3.0. The README also offers a separate commercial license |
| Architecture | A planner and 3 workers by default over an asset graph and an exploration graph of goals, intents, facts, findings and hints | A blackboard graph of facts, intents and hints. A dispatcher runs workers that reason or explore, with no fixed roles |
| Agent runtime | Its own workers, calling any Anthropic or OpenAI compatible endpoint | Workers are Claude Code, Codex or Pi command-line agents |
| Guard and safety controls | Every tool call passes intercept rules, an LLM judge and human approval. A recording proxy keeps all traffic | People can add hints at any time. Workers run in a container per project, or in a local mode the README says has no sandbox. No approval step described |
| Deployment | One Go binary with a web console and PostgreSQL, from release zips or Docker | Docker Compose with a server on port 8000 and a dispatcher, or a local mode without Docker |
| Maintenance | RuoJi6/ARTEX: v0.3.30 on 2026-11-14, last push 2026-11-14, 2,194 stars (2026-11-17) | v0.2.1 on 2026-05-10, last push 2026-09-07, 5,310 stars and 1,090 forks (2026-11-17) |
| Offsec Harness Index | Entry 01, tier Confirmed, breach stated. 14 primary reports. Key figure: 200+ organizations targeted (CrowdStrike) | Tier Confirmed, breach stated. 6 primary reports, including Unit 42, Gambit and NorForge. Key figure: 27+ companies compromised in one campaign |
| Public benchmark | Baidu Agent+ final: 1st overall (2026-09-03). Live range 67.83/100, 7th. TSecBench 89.78/100, 6th | README: 54 of 54 problems solved and 3rd place at the Tencent Cloud Hackathon AI pentest challenge, 2nd edition |
A shared idea
Cairn's README describes a blackboard: a shared graph that grows from an origin toward a goal. A fact is a confirmed finding, an intent is a direction not yet explored, and a hint is human judgment added at any time. Workers read the whole graph, decide what to explore next, and write new facts back. They talk only through the board. The README lists three task types: bootstrap, reason and explore.
ARTEX uses the same words. Its exploration graph holds goals, intents, facts, findings and hints, workers claim one intent and write facts back, and a planner reads the graph to decide what comes next. The RuoJi6/ARTEX README has a short reference section that links only to oritera/Cairn. ARTEX adds a second, shared asset graph of domains, hosts, services and endpoints, a console, and a guard on every tool call.
Runtime and controls
Cairn does not ship its own agent. Its workers are Claude Code, Codex or Pi command-line agents that receive a prompt and return structured output. By default each project gets a worker container. In local mode the agents run on the host with the user's own permissions, and the README says plainly that there is no sandbox. Human control comes through hints. The README does not describe a step where a person approves each action.
ARTEX runs its own workers against a model endpoint you set, and puts each tool call through intercept rules, an LLM judge and human approval, with a recording proxy for evidence. That makes ARTEX the heavier install, with PostgreSQL and a console, and Cairn the lighter one.
In the Offsec Harness Index
Both tools have the tier Confirmed in the Offsec Harness Index. For ARTEX it cites 14 primary reports, and its key figure, 200+ organizations targeted, comes from CrowdStrike's report of 2026-10-29.
For Cairn it cites six primary reports. Unit 42's report of 2026-11-12 says Cairn confirmed unauthorized access or data retrieval at 10 targeted organizations. Gambit Security's report of 2026-09-22 counted at least 27 companies compromised in one campaign against online retailers, and NorForge's report of 2026-10-19 describes exposed Cairn panels. Google and CERT-In also name it. JPCERT/CC listed Cairn beside ARTEX in its Japan alert. These rows record what was reported. They do not rank the tools and say nothing about the maintainers' intent.
Status, license and numbers
Both are AGPL-3.0. Cairn's README also offers a commercial license for use without the AGPL terms, and contributors agree their code may be used under both. Cairn's last release, v0.2.1, came on 2026-05-10 and its last push on 2026-09-07. RuoJi6/ARTEX released v0.3.30 on 2026-11-14.
Cairn's README reports 54 of 54 problems solved and 3rd place at the 2nd Tencent Cloud Hackathon AI pentest challenge, among 610 teams. ARTEX scored 89.78 out of 100 on the Tencent TSecBench blind range and was first overall at the Baidu Agent+ final. These are different contests with different scoring, and Cairn's result is self-reported in its README, so this page does not set one against the other.
Sources
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 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 AgenticPentest
ARTEX is AGPL-3.0 and gates every tool call; AgenticPentest is a two-week-old MIT workbench that fences work by scope and approves state-changing actions.
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.
