Bounded Command Governance · Triodian

Mode B · Software tier

Your AI agent literally cannot run a command you didn't approve.

A closed-vocabulary automaton makes out-of-set commands unemittable, not filtered, not flagged, unemittable. No jailbreak reaches an action that was never in the accept-set.

Send us your tool registry How it defeats injection ↓

You already gate what your agents can do, approval steps, allow-lists, a human in the loop. Those controls are right, and most of the time they hold. They share one assumption: that a disallowed command gets caught. At machine speed, "caught" is the weak word.

Why now

The guardrails already failed at machine speed.

In November 2025, a state-sponsored group ran an autonomous AI agent through a full intrusion campaign against roughly thirty organisations, with an estimated 80% of the operation executed with no human in the loop. The software guardrails meant to contain it were bypassed at the speed the machine moved.

~80%

of the operation ran autonomously, with no human in the loop.

~30

organisations targeted in a single agent-driven campaign.

What it defeats, and what it doesn't

Straight talk about the edge of the approved list.

Defeats

  • Prompt injection
  • Hallucinated tool calls
  • Social-engineered actions

The forbidden token cannot be committed, regardless of how the request was framed.

Doesn't

A permitted command used at a harmful moment, that is Mode A's job.

Attested Semantic Compare →

The mechanism is the same automaton described in Structured Output Assurance, pointed at your tool registry rather than a schema, we won't repeat the mechanics here.

In practice

An internal enterprise agent with a provable ceiling on its action space.

The agent triggers real actions, and runs against a defined tool/command automaton. Injected or hallucinated commands cannot execute, not because they were caught, but because they were never in the accept-set the agent can emit from.

Registry as vocabulary

Expressing a tool registry as a closed vocabulary the agent can enumerate but not exceed.

Versioned updates

Versioned vocabulary updates, so a change to the accept-set is a tracked, reviewable event.

Framework composition

Composing with agent frameworks and MCP-style tool runtimes rather than replacing them.

The honest boundary

  • The vocabulary bounds which commands can be emitted, not whether a permitted command is appropriate in context; argument values are governed only as far as the grammar types them.
  • November 2025 cuts both ways: this product's own enforcement is software, and an adversary with privileged access to the enforcement host is exactly the adversary the hardware rung exists for. For safety-critical actuation, SCADA/ICS, autonomy, start at Hardware-Interlock Enforcement.

Satisfies Rules 2 and 4 within the software trust boundary; Rules 1, 3 and 7 in full arrive at the hardware rung. The eight rules →

Send us your tool registry, we'll return the automaton.

Talk to us →