Kryon v2.1.0 — Hydra · Skillforge

First public release of Kryon — a local-first, autonomous offensive-security agent.

Point it at a target and, from a single prompt, it runs recon, authorized pentesting, vuln hunting, DFIR, or a compliance audit — then hands back findings and a report. It runs fully on your own hardware via a local, OpenAI-compatible LLM (llama.cpp), so there's zero API cost and no data leaves the engagement perimeter.

Highlights

• Skill-based architecture — one unified agent + ~110 markdown playbooks matched by target profile and operator intent.
• Deterministic-first — critical detectors (nuclei, nikto, sqlmap + hybrid probes) run as pre-hooks before the LLM: the model narrates evidence, it can't skip the scanner.
• Multi-framework compliance — PCI-DSS, CIS, NIST CSF, SOC 2, ISO 27001, HIPAA, GDPR, PSD2/FAPI, plus 20+ production-capable audit modules (FortiGate, Proxmox, Windows, Tomcat, ESXi, KVM, Hyper-V, Xen, …).
• Grows without a bigger model — kryon update refreshes the nuclei / CVE / ExploitDB detection feeds.
• Banca-safe by default — passive recon, throttled scanning; intrusive tools gated behind KRYON_RED_TEAM.
• CI/CD integration — reusable GitHub Action + GitLab CI template to run Kryon as a SARIF security gate.

Install

See the README for Docker deployment. The Python package (wheel + sdist) is attached to this release.

License

Licensed under AGPL-3.0-or-later. See LICENSE and NOTICE for third-party attributions.