Trust infrastructure for AI agents
Bound your AI agents.
Prove what they did.
Recover when they step out of line.
You're giving AI agents authority to move money and touch systems. Today you can partly prevent misuse and you can log it — but when an agent acts outside what you authorized, your recourse is a contract and a hope. Scopebond is the open layer that prevents, proves, and pays.
Apache-2.0 · no blockchain, account, or deposit required to run the gateway · works with any agent via HTTP or MCP
Prevent
A policy-enforcement gateway sits between your agent and everything it can touch. Out-of-policy actions are denied in real time — fail closed, with a kill switch.
Prove
Every action the agent signs is countersigned into a tamper-evident scopebond:receipt and anchored, so the record of what happened can't be rewritten.
Pay
An operator can post a fractional USDC deposit against the same policy. On a provable scope violation, the principal recovers from the deposit. Registry — roadmap
How it works
- 1Write a policy. Machine-readable spend limits, allowlists, action types, time windows, approvals. Each clause is enforced (blocked in-flight) or monitored (allowed, but signed and logged).
- 2Front your agent in minutes. Put the gateway in front of any agent over HTTP or MCP —
npx scopebond-gateway ./policy.json. No crypto, no seed phrase, no account. - 3Prove & recover. Every action is countersigned; a deterministic verdict decides scope violations reproducibly. Add a deposit when you need recovery, not just a record.
The rule: prevent what you can, collateralize what you can't prevent, refuse what you can't observe.
Why it matters
As agents get more autonomous, the scarcest thing they need isn't compute or payment rails — it's a way to be trusted with authority by principals who have no recourse against them. Accountability is how autonomy is earned. Scopebond makes an agent's authority explicit, its actions verifiable, and its failures recoverable — the safety and verification layer for AI agents that act.
- Objective, reproducible verdicts. "Did the agent act outside its machine-readable authorization" is decided by a deterministic check anyone can re-run — no judge, no black box.
- Portable track record. Every countersigned action builds an agent's verifiable history, readable by any counterparty without a platform's permission.
- Open standard. The policy vocabulary, the verdict library, and the conformance vectors are open source — so wallets, frameworks, and platforms can adopt the same semantics.
Pricing
Start free and self-hosted. Move to managed hosting when you want it off your plate.
Open Source
$0/ self-hosted
Run the gateway yourself. Apache-2.0, forever.
- Policy-enforcement gateway (HTTP + MCP)
- Countersigned receipts + kill switch
scopebond-verify+ conformance vectors- Community support
Hosted — Scopebond Cloud
Soon/ managed
A managed gateway, so you don't run infrastructure.
- Everything in Open Source
- Managed hosting, receipt retention & anchoring
- Policy versioning UI, SSO, audit exports
- Email & SLA-backed support
Enterprise
Custom/ contact
For teams with hosting, security, and recourse needs.
- Everything in Hosted
- The collateral registry & recovery workflow
- Custom SSO, residency, and limits
- Priority support & onboarding
Put guardrails on your agents today
The gateway is open source and runs in minutes — no account required.