Autonomous security
intelligence in your terminal
RedCell is an AI-powered security agent built for authorized penetration testing, bug-bounty recon, and vulnerability discovery. We combine high-reasoning AI models with strict scope enforcement to automate complex security workflows from the command line.
Our Mission
Security teams and researchers spend hours manually stitching together recon tools, running scanners, filtering false positives, and writing reports. RedCell was created to replace repetitive security toil with an autonomous terminal agent that plans, executes, verifies, and documents engagements — safely and reproducibly.
Built on four core principles
Engineering choices designed for credibility, safety, and speed.
Terminal-First Ergonomics
RedCell operates where security professionals spend their lives — in the shell. A single compiled binary with zero external runtimes or Python dependencies.
Zero-Trust Scope Guarding
Every action is intercepted by an active policy engine that verifies target authorization, rate limits, and forbidden actions before sending any network traffic.
Model Context Protocol (MCP)
Plug your custom exploit scripts, Python tools, and third-party security servers directly into RedCell using open MCP standards.
Evidence-First Reporting
Every finding is automatically captured with proof-of-concept verification steps, severity scores, and raw log artifacts into exportable Markdown & JSON.
How RedCell works under the hood
A decoupled, safe execution loop between host binary and AI inference.
Host-Native Binary
Runs directly on your host machine (macOS, Linux, Windows). Local state stays local in ~/.redcell/.
Scope Policy Engine
Interprets rules of engagement and validates target IP/domain bounds before any tool call execution.
AI Multi-Model Backend
Utilizes high-reasoning LLMs via secure proxies to plan multi-stage recon, scan, and verification sequences.
Extensible Tool Execution
Orchestrates web-nuclei templates, passive OSINT scripts, local Tor proxies, and custom registered MCP servers.
Values & Rules of Engagement
Strict Authorization Only
RedCell is designed exclusively for authorized penetration testing, security auditing, and bug bounty hunting inside explicitly permitted target boundaries.
No Hallucinated Vulnerabilities
Unconfirmed LLM outputs are treated as hypotheses, not facts. RedCell runs real proof-of-concept verification checks to eliminate false positives.
Privacy & Local State
Your investigation logs, discovered credentials, and engagement sessions remain stored on your host in ~/.redcell/sessions/.
Open Ecosystem Interoperability
We support open standards like MCP and open-source nuclei templates so security teams can bring their own custom security tooling.
Security & Compliance
We take security seriously. Read our security policies, acceptable use terms, and vulnerability reporting procedures.
Get in Touch
Have questions, enterprise requirements, or security inquiries? Reach out to our engineering team.
Experience RedCell in your terminal
Single compiled binary. Free plan available. One install command.