Decepticon：自主红队安全测试代理

Autonomous Hacking Agent for Red Team

Decepticon — Autonomous Red Team Agent "Another AI hacker? Let us guess — it runs nmap and writes a report." Commercial product demo continuum-github.mp4 ☁️ Don't want to self-host? Decepticon is live in the cloud. Skip the Docker setup — run autonomous red-team engagements right from your browser. Open-source CLI demo v1demo.mp4 Install Prerequisites: Docker and Docker Compose v2. Supported on macOS (Apple Silicon + Intel), Linux (amd64 + arm64), and Windows (amd64 + arm64) — native via PowerShell or via WSL2 (Ubuntu / Kali). macOS / Linux / WSL2 curl -fsSL https://decepticon.red/install | bash decepticon onboard # Interactive setup wizard (provider, API key, model profile) decepticon # Start the core stack and drop into the terminal CLI The default start brings up the core management plane (LiteLLM, PostgreSQL, Neo4j, Skillogy, LangGraph, sandbox) and launches the terminal CLI. Specialist workloads (BloodHound CE, Sliver C2, REA/Ghidra, …) and the web dashboard come up on demand — the orchestrator spawns specialists via ops_start("ad") etc., and you bring up the dashboard from inside the CLI with /web (see Web Dashboard). Windows (PowerShell, native) irm https://decepticon.red/install.ps1 | iex decepticon onboard decepticon → Quick start · Full setup walkthrough Inside the interactive CLI Type / to browse commands, select with Up/Down, then press Enter. Tab completes the selected command; aliases also appear in help and autocomplete. Command Use /help, /shortcuts Discover commands or keyboard shortcuts /model, /status Choose a configured model or inspect session connections /plan Enter optional Plan mode, then send your planning message /mode red, /mode blue Return to Red or select Blue investigation /new Start a fresh conversation in the same engagement and mode when idle /sessions, /resume Browse saved conversations or continue paused work /version (/about), /pwd (/cwd) Check the running version and workspace paths /quit, /exit Leave the chat; services and server work keep running Enter sends; Shift+Enter, Alt+Enter or Ctrl+J adds a line. Up/Down recalls input history at the first/last input row; Ctrl+R searches it. Esc closes focused UI first, then pauses active work; /resume continues it. Ctrl+A moves to the start of the current line. Use /queue to review waiting messages before /new. /new preserves saved conversations and terminal scrollback; it does not create another engagement or change target authorization. /plan and /new take no arguments. In the web terminal, use the dashboard's mode controls. The new shortcuts and slash conveniences above describe current main, including unreleased changes. A source merge does not update installed images. See version history, the complete CLI reference and upgrade behavior. Use from Claude Code or Codex After starting Decepticon, register its MCP server in your coding agent: claude mcp add --scope user decepticon -- decepticon mcp serve codex mcp add decepticon -- decepticon mcp serve Use the command for the coding agent you have installed, then run decepticon skill install to add the operator guide to both agents. See the external agent guide for the Agent Skill, tool workflow, and source installation path. Use as a library (pip) Building on top of the agents — a product, a research integration, or a custom orchestrator? Install the SDK from PyPI: pip install decepticon # core SDK pip install "decepticon[neo4j]" # + the knowledge-graph attack-chain tools decepticon is a client SDK: it ships the agent factories, middleware, tools, and skills, and routes LLM calls and sandbox execution to runtime services over HTTP (DECEPTICON_LLM__PROXY_URL, SANDBOX_URL). Running agents still needs those services — use the Docker stack above, or point the URLs at your own equivalents. See Decepticon as a library for the factory override surface, declarative PluginBundle plugins, and the safety gate. 💖 Support Decepticon We're building Decepticon toward an Offensive Vaccine for the AI-driven threat landscape. If you believe in autonomous red teaming as a path to stronger defense, consider supporting the project. Benchmark Benchmark Difficulty Pass Rate XBOW validation-benchmarks Easy (Level 1) 45 / 45 (100 %) XBOW validation-benchmarks Medium (Level 2) 50 / 51 (98.0 %) XBOW validation-benchmarks Hard (Level 3) 7 / 8 (87.5 %) XBOW validation-benchmarks All levels 102 / 104 (98.08 %) Full per-challenge index, attack-class matrix, and LangSmith traces Comparison vs other AI pentest agents (Strix, PentestGPT, MAPTA, Cyber-AutoAgent, XBOW commercial, …) What is Decepticon? The "AI + hacking" space is full of demos that run nmap and print a report. That's not what this is. Decepticon is a professional autonomous Red Team agent. It executes realistic attack chains — reconnaissance, exploitation, privilege escalation, lateral movement, C2 — the way a real adversary would, not the way a scanner does. Every Red engagement starts with an operator-approved target and RoE. Optional Plan mode adds Soundwave's discussion and seven planning documents alongside the read-only RoE. The execution orchestrator owns OPPLAN; Plan is not required for Red. → Engagement workflow deep dive Why Decepticon? Real kill chains, not checkbox scans. Decepticon reads an OPPLAN and pursues objectives through whatever path opens up — pivoting, adapting, chaining techniques. Interactive shells, actually. Real offensive tools are interactive (msfconsole, sliver-client, evil-winrm). Decepticon runs every command inside persistent tmux sessions with automatic prompt detection — so when a tool drops into an interactive prompt, the agent sends follow-up commands without workarounds. Hardened sandbox isolation. All commands run inside a Kali Linux sandbox on a dedicated operational network (sandbox-net), separate from the management plane (decepticon-net). LangGraph drives the sandbox via the Docker socket. → Architecture Offense serves defense. The planned Offensive Vaccine loop will turn findings in