Decepticon — Red Team Operations Powered by AI Agents

Decepticon — Red Team Operations Powered by AI Agents

A framework for automating red team operations, from reconnaissance and vulnerability research to exploitation. The operator specifies an authorized target, and the agents carry out tasks in an isolated environment. Decepticon uses 16 specialized agents, each with a fresh context for its objective, coordinated by an orchestrator.

Key features: • Attack Planning — optional planning mode generates an operational plan mapped to MITRE ATT&CK techniques. • Interactive Tools — maintains persistent sessions for tools such as Metasploit, Sliver, and Evil-WinRM, allowing agents to continue working in interactive consoles. • Persistent Findings — stores findings in project files and a knowledge graph, allowing agents to build on results from previous stages even when each task starts with a fresh context.

Decepticon can be used through a command-line interface or web dashboard, and integrated with Claude Code and Codex via an MCP server. Self-hosted deployment requires Docker, Docker Compose v2, and access to a supported language model provider.

Tool: https://github.com/BitterSecurity/Decepticon