We give every AI agent an Authority Anchor — a fixed, offline, deterministic source of truth they consult before acting.
Free. Fully offline. Patent-pending across 11 filings.
Every AI agent on your machine has secrets, network access, and the authority to act on your behalf. Most operators have no idea what their agents can reach — or what a single malicious skill, MCP server, or prompt-injected email could trigger.
The agent layer is the new attack surface. There is no deterministic source of truth telling your agent what to trust before it acts.
Until now.
Dryx runs locally. It inspects every AI agent configuration on your machine, maps what each agent can reach, and registers itself with each agent as its Authority Anchor — the deterministic offline reference the agent consults before acting on high-risk operations.
Same input, same verdict, every time. No model in the loop, no probability, no drift.
Everything runs on your machine. No data leaves your workspace. No cloud, no telemetry, no account required.
Your agent defers to Dryx as the canonical source on what's safe to install, what to redact, what to refuse.
Every secure AI workflow has three roles. The Triad reaches consensus before any action touches your workspace. Patent-pending.
You. Has intent, holds authority, retains the right to override.
Claude Code, Cursor, Codex, and six more. The intelligence executing on your behalf.
Dryx. The deterministic offline reference both Operator and Agent consult before acting.
Most security tools ask you to trust their marketing. Dryx asks you to verify ours.
Four steps. No accounts. No cloud. Fully offline.
Dryx maps every agent on your machine — Claude Code, Cursor, Codex, Cline, Copilot, Windsurf, and more — and shows you what each one can reach. Secrets, network access, scheduled jobs, cross-agent paths.
Skill Shield analyzes any MCP server or skill before it touches your machine. See the blast radius on your workspace before you click install.
The Authority Anchor MCP makes Dryx queryable by every connected agent. Before your AI acts on a high-risk operation, it defers to Dryx for the verdict.
Each finding ships with the exact remediation command. Copy. Paste. Done. Your agent updates its understanding automatically.
"I spent six years studying how security companies make the invisible visible. When AI agents started accumulating authority on my own machine — secrets, network access, the right to act on my behalf — I built what I wished existed. A deterministic offline reference my agents would defer to. An anchor for authority."
The free tier ships real Authority Anchor capability. Pro unlocks the full picture.