Offensive security tooling and services. We build open-source AI-driven pentesting infrastructure and run security engagements against production systems.
Backed by Techstars and Outlier Ventures.
pentest-swarm-ai — Autonomous penetration testing using a swarm of AI agents. Coordinates recon, classification, exploitation, and reporting through a stigmergic blackboard. Go, Claude API, 7+ native security tools. AGPL-3.0.
agat — Fingerprints polymorphic and metamorphic malware that mutates between runs. Behavioral analysis pipeline that survives encoding, packing, and code rewriting. Featured at BSides Bangalore 2026.
akira — [one-line description here]
skillscan — [one-line description here]
vibescan — AI-driven security scanner for vulnerabilities in AI-generated code. Combines agents with semgrep, nuclei, trivy, bandit, gitleaks, and snyk. Go, MIT.
We run penetration tests against web applications, cloud infrastructure, and internal networks. Same toolchain we open-source, executed by the team that built it.
For engagements: akhil@armur.ai
Our team publishes on AI security and adversarial systems. Recent work includes XAMT: Adversarial Attacks on Multi-Agent Systems (arXiv:2512.15790).
- Web — armur.ai
- Engagements — akhil@armur.ai
- X — @armur_ai
- LinkedIn — armurai