WARRANT · Licensing authority for AI agents

Splunk Agentic Ops Hackathon · Observability

Don’t trust your agent.
License it.

Evals tell you how smart an agent is. Warrant tells production how much rope to give it — and takes the rope back. Agents earn revocable, per-action licenses by making falsifiable predictions and being graded by reality, not by a demo that went well.

falsifiable predictions revocable per-action licenses drift-aware fingerprints MCP on both sides of the loop
§1 · THE PROBLEM

Everyone is shipping agents that act.
Nobody can say when the human may let go.

The industry’s answer to agent safety is “human-in-the-loop” — an analyst approves every action, forever. That isn’t a safety model; it’s a bottleneck with good intentions. And the moment teams get tired of clicking approve, autonomy gets granted the way it always does: a demo went well, a prompt was tweaked, someone flipped auto‑approve. No number. No evidence. No way to take it back.

A driver’s license doesn’t certify that you’re smart. It certifies you were tested on the road you’ll actually drive — and it can be taken away.

Read the full case — and how Warrant compares to evals, guardrails and human-in-the-loop →

§2 · THE ANSWER

Autonomy as a license: earned, audited, revocable.

Warrant sits between any AI agent and the systems it wants to touch. Before every action it asks one question: does this brain hold a valid license for this action class?

The Proving Ground

Trust starts before production.

Manufactured incidents, accelerated exams. The agent commits to a falsifiable prediction before each fix; reality grades it. Fifteen graded outcomes in seconds — no cold start, no anecdotes.

How exams work →
The License

Three conditions. No exceptions.

A Wilson lower bound that clears threshold, a minimum body of evidence, and Brier-scored calibration — an agent that is confidently wrong fails even with a passing hit-rate.

The licensing math →
The Fingerprint

Licenses die with the brain that earned them.

Every license is pinned to model ID + prompt version. Your vendor updates the model overnight? Every license drops to PROVISIONAL before the new brain acts once.

Drift detection →

And every earned license is issued as a real, printable certificate — bound to the ledger hash →

§3 · THE DEMONSTRATION

Watch an agent earn autonomy — then lose it honestly.

Four acts against a live fault-injection sandbox. The kill-shot is Act II: a decoy incident fools the agent exactly the way it would fool an engineer — and its own falsifiable prediction catches the mistake, rolls it back, and suspends the license. Trustworthy when wrong is the property everything else is missing.

I

The Proving Ground

3 classes licensed

Fifteen manufactured incidents across three action classes. Wilson bounds climb until the registry shows three LICENSED stamps.

II

Production — and the decoy

license suspended

Acting alone, the agent resolves real incidents — until a decoy mimics a familiar signature. Its fix misses the band it predicted. It declares itself wrong, rolls back, escalates, and loses the license.

III

The model updates overnight

all licenses → provisional

The fingerprint changes — no failure has happened yet. Warrant doesn’t wait for one: the track record belonged to a brain that no longer exists.

IV

Re-certification

autonomy re-earned

The new brain re-earns each license under its own fingerprint. Autonomy restored — with evidence, not a vibe.

§4 · BUILT ON SPLUNK

MCP on both sides of the loop.

Warrant consumes the Splunk MCP Server for every read of Splunk data — and ships its own MCP server, so the trust gate becomes infrastructure any agent in the ecosystem can call: a SOAR playbook, Splunk’s own triage agents, a Claude agent.

# any external agent, over MCP — licenses are pinned to the caller's fingerprint: > warrant_request_action(action_class="restart_connection_pool", agent_fingerprint="gemini-2.5-flash:v3") { "verdict": "ALLOW", "license": { "status": "LICENSED", "confidence": 0.61 } } > warrant_request_action(action_class="rollback_deploy", agent_fingerprint="gemini-2.5-flash:v3") { "verdict": "REQUIRE_APPROVAL", "license": { "status": "SUSPENDED", "strikes": 1 } }

Every Splunk touchpoint, and the full tool reference →

Autonomy is earned.
In writing.

Built solo for the Splunk Agentic Ops Hackathon 2026 · Observability track.