[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"$ffwppi573btvf":3},{"code":4,"message":5,"data":6,"timestamp":26},10000,"success",{"full_name":7,"position":8,"description":9,"language":10,"stars":11,"forks":12,"stars_gained":13,"age_days":14,"hn_threads":15,"one_liner":16,"summary_md":17,"topic":18,"github_topics":21,"license":22,"homepage":23,"updated_at":24,"history":25},"azrielnan\u002Fartex-ko-main",20,"面向已获授权安全测试与本地研究的 AI 多 Agent 安全测试平台。","Go",116,5,11,10,[],"A web platform that coordinates AI agents for authorized security testing and local reverse engineering, keeping findings, traffic and evidence in one place.","ARTEX is worth a look if you already run authorized penetration tests and want an agent-driven workflow with an auditable trail; it is not a turnkey scanner.\n\n## The problem\nAn authorized test typically starts with asset discovery, then manual probing, then notes scattered across terminal logs, a proxy history and a findings doc. The tester has to stitch those together before writing a report, and evidence for each finding often has to be reconstructed afterward. ARTEX shifts that workflow into one web platform: a planner proposes steps, workers execute tool calls, and assets, exploration paths, recorded traffic and findings are persisted with their sources. The result is a task record you can pause, resume and turn into a report instead of a pile of separate artifacts.\n\n## What's different\nThe design centers on orchestration rather than a new scanner. A Go backend embeds a Next.js frontend into a single binary, uses PostgreSQL for persistence, and has agents invoke tools through the `norma` SDK, recording traceable evidence. A recording proxy captures requests, responses and activity paths, and vulnerabilities get their own retest sessions that preserve conclusions and history. Human-in-the-loop approval is built into task flow, so you can step in before an agent continues.\n\nSecurity testing and reverse engineering are deliberately separated: separate sessions, prompts, agent context, assets, traffic and findings, so a reverse-engineering task cannot read security-testing state. Reverse engineering is optional and off by default, using the REA MCP server for local binaries, packages, source trees and runtime material [REA project](https:\u002F\u002Fgithub.com\u002Fmorluto\u002Frea).\n\nThe trade-off is operational surface area. You are running a Go server, a PostgreSQL instance and a recording proxy, and optionally a Node-based REA runtime, so there is more to configure and keep healthy than a command-line tool. The README also notes that the default Docker Compose file pulls the upstream `autumn27\u002Fartex` image, which may lag this fork's source, so the compose path is not the way to get the newest features [project instructions](https:\u002F\u002Fgithub.com\u002Fazrielnan\u002Fartex-ko-main\u002Fblob\u002Fmain\u002FREADME.md). A public demo exists with clearly labeled mock data and no real targets [online demo](https:\u002F\u002Fazrielnan.github.io\u002Fartex-ko-main\u002F).\n\n## Who should try it\nGood fit: a small security team or independent tester who runs authorized engagements and wants agent-assisted discovery plus a persistent, reviewable record of what was touched and what was found. The approval step and separated modes are useful when results need to be explained to a client or reviewed internally.\n\nPoor fit: someone looking for a one-command scanner, or anyone testing systems without written authorization. The README's disclaimer is explicit about legal responsibility, and the platform is not a substitute for scope decisions. Also a poor fit if you only need reverse engineering: the REA mode requires enabling an MCP server and, outside the release package, Node.js 22.19+, 24.11+ or 26+ plus each analyzer's own dependencies.\n\n## Requirements and limitations\nFour setup paths exist, and they differ in what they demand. The install script needs Docker or local Go, Node.js and PostgreSQL; Docker Compose needs only Docker plus an edited `.env`, but may serve older code; the Windows Release package bundles PostgreSQL and the REA runtime and needs no Go, Node.js or PostgreSQL installed, though it must be fully extracted to a writable directory and started via the launcher scripts rather than the raw executable [release packages](https:\u002F\u002Fgithub.com\u002Fazrielnan\u002Fartex-ko-main\u002Freleases). Linux and macOS require a preconfigured PostgreSQL.\n\nOngoing upkeep matters more than first install. The launcher scripts handle restart after updates and rollback on failure, and the README warns against running the binary directly instead. Windows users upgrading from 0.3.19 need the full package to update old launcher scripts, and should back up data and avoid mixing PostgreSQL major versions. LLM access requires your own provider, model and API key, configured in system settings or via environment variables.\n\n## HotRepo's take\nHotRepo first tracked the repository on 2026-10-24, when it was 10 days old, and recorded it at #20 on the NEW-REPOSITORY board (新项目榜, not the overall Trending board) on 2026-10-25 with 116 total stars; that one-date sample is not a daily growth series. Trend persistence cannot be judged from a single dated observation. The practical point to verify before investing time is whether the setup path you intend to use actually includes the features you want, since the README itself flags that the default Compose image may trail the source.",{"slug":19,"name":20},"security","Security & Privacy",[],"AGPL-3.0",null,"2026-10-25T06:47:42Z",[],1792919780]