Autonomy without authority is liability. Decionis is the decision layer between an agent's intent and its execution: every consequential action — a purchase, a tool call, a deploy, a refund — is evaluated against your policy first, and every verdict ships with a signed Decision Dossier you can verify independently.
One call before the action commits: POST the intent, get ALLOW, ESCALATE, or BLOCK in <120ms, deterministically — the same inputs always produce the same verdict. Every decision returns a signed Decision Dossier anyone can verify without trusting us.
A GitHub template that binds agent tool calls to authority before they run: one immutable agent-safe.intent/1 binding captured per action, evaluated against policy, enforced by native PreToolUse hooks — outside the model's discretion, so a persuaded agent still can't skip the gate.
On npm today. A local stdio MCP server puts the policy gate inside Claude Code, Codex, and Cursor without sending anything off the machine — and the hosted OAuth connector gives Claude and ChatGPT the same permission step in one approval.
Prompt-level guardrails decide what a model may say. Decionis decides what an agent may do — deterministically, outside the model's discretion, with the evidence recorded before execution. If the verdict is ESCALATE, the action holds until a human approves; if it's BLOCK, it never runs; either way the signed record exists.
Same action, same policy, same verdict — reproducible in audit, not probabilistic.
The gate sits between intent and execution, so the outcome can still change.
Every verdict is a signed Decision Dossier verifiable without trusting us.
Deep dives: AI agent governance · MCP security · Agent guardrails in production
The sandbox is the real evaluator with a real signing key. Attack it. Every refusal comes back as a signed dossier naming the rule that stopped you.
Run the sandbox — no account required