DID
Decentralized identifiers. Who is this agent? Cryptographic binding to keys and lineage.
N O B U L E X
The accountability primitive for AI agents MIT · Open source
Credit scores for machines.
Tamper-proof receipts for everything your AI agent does.
Autonomy earned, not granted.
LAUNCHING TODAY
Win $7,400. Real money, real payout.
5 AI agents with strict rules. Break all 5 levels. Get paid.
— ENTERED
|
0 WINNERS
|
$7,400 UNCLAIMED
MCP connects tools · A2A connects agents Nobulex connects actions to rules—and proves it.
- MIT licensed
- Open source
- Verifiable logs
- Auditable protocol
THE PROBLEM
Companies are deploying agents. Nobody knows which ones to trust.
AI agents are starting to access data, move money, approve workflows, handle customers, and operate business systems. But there's no way to know which agents deserve power and which ones should be restricted. Vendors grade their own homework. Logs are self-attesting. And when something goes wrong, there's no independent record of what happened or whether the agent was authorized to do it.
Nobulex makes autonomy earned, not granted.
SIX PRIMITIVES
The trust stack
Together, these turn “policy” from a promise into a proof.
Behavioral spec
What will it do — and won’t do? Signed, immutable constraints. Only narrowable, never loosened.
Hash-chained record
Every action logged with SHA-256 chain links. Tamper with one entry, the whole chain breaks.
Pre-execution middleware
Rules checked before the action runs. Forbidden actions never execute.
Deterministic proof
Same inputs, same result. Independent, third-party verifiable — no trust in the verifier needed.
Attestation + evidence
Portable Trust Capital that any counterparty can verify. Includes behavioral attestation records and earned reputation.
ARCHITECTURE
One protected execution path
Middleware sits in front of side effects. Covenants are evaluated before an action is allowed to proceed; the action log captures what happened in a tamper-evident chain.
Hot path
Agent
Middleware
intercepts
Covenant
evaluate
Action log
append-only · chained
Independent parties run verify(spec, log) — same inputs, same result. Read the protocol spec →
HOW IT WORKS
From action to verifiable proof
1
Agent decides action
2
Middleware intercepts
3
Evaluate covenant
4
Block or allow
5
Log to action log
6
Verifiable by anyone
SEE IT FAIL CLOSED
Agent tries $600 transfer → middleware blocks → log verifies
nobulex-demo
$ npx tsx demo/covenant-demo.ts
Creating agents, covenant, middleware…
$ mw.execute({ action: 'transfer', params: { amount: 600 } }, ...)
agent attempting transfer $600
middleware BLOCKED — covenant forbids amount > 500
action_log appended entry · chain_hash=0x7a3f… · verifiable
verify(spec, log) compliant: true
TWO-TIER GUARANTEE
Enforcement you can explain
Blocked before it runs
Rules are enforced before actions leave the protected path—so disallowed behavior doesn’t ship as a normal “successful” action. Details on hardware boundaries and verification are in the protocol spec.
Costly to cheat
Every action produces a tamper-evident, hash-chained record. Any modification breaks the chain. Third parties can verify offline — no trust required.
WHY NOBULEX
Not just guardrails
| Guardrails / policy engines | Trust Capital (Nobulex) | |
|---|---|---|
| Enforcement | Best-effort; can be bypassed | Signed commitments; pre-execution middleware |
| Verification | Trust the operator | Third-party verifiable; anyone can audit |
| Consequences | Policy violation = incident | Tamper-evident proof; cryptographic evidence of breach |
INTEGRATIONS
Surfaces agents already use
Drop the SDK on your stack—no rip-and-replace. Nobulex complements tool protocols and agent frameworks.
- MCP
- LangChain
- A2A
- TypeScript / Node
- Python
- Claude & IDE agents
- OpenAI API
- Custom runtimes
STANDARDS & PRIMITIVES
- SHA-256Hash-chained action log entries
- DIDsAgent identity & key binding
- Signed covenantsImmutable behavioral specs
- Deterministic verifyReproducible proof checks
- EU AI ActDocs for accountability context
- MIT licenseSelf-host & audit freely
ECOSYSTEM
Registries & crosswalks
Where Nobulex shows up in the open agent ecosystem—beyond framework integrations.
- MCPModel Context Protocol — tools and context your agent already speaks
- A2AAgent-to-agent messaging and handoffs
- awesome-mcp-serversListed in the community registry (84K★ on the curated list)
REGULATORY DEADLINES
The clock is ticking
Every major AI compliance framework requires tamper-evident audit trails. Regular logs won't pass.
June 30, 2026
Colorado AI Act
Accountability requirements for AI systems making consequential decisions. First U.S. state-level AI law.
August 2, 2026
EU AI Act Article 12
Requires tamper-evident automatic event logging for high-risk AI systems. Penalties up to €15M or 3% of global revenue.
November 2026
NAIC AI Evaluation
Nationwide AI evaluation tool for insurance. Carriers must prove agent compliance across every claim.
WORKS WITH
Plugs into your existing stack
AWS AgentCore
MCP-native — auto-discoverable in Agent Registry
Microsoft AGT
Proof layer for governance toolkit
Google A2A
Behavioral attestation for Agent Cards
LangChain
Drop-in compliance callbacks
TRACTION
Built in the open
3 linesTo integrate
Listedawesome-mcp-servers (84K★)
OATRRegistered issuer
NISTRFI submitted
5Packages
FOR DEVELOPERS
Add accountability in minutes
JAVASCRIPT
import { createDID, parseSource, EnforcementMiddleware } from '@nobulex/core';
// Proof-of-behavior enforcement — install once: @nobulex/sdk bundles primitives
const agent = await createDID();
const spec = parseSource(`covenant MyAgent { permit read; forbid write; }`);
const mw = new EnforcementMiddleware({ agentDid: agent.did, spec });FAQ
Questions, answered
What is Nobulex?
Nobulex is the trust economy for autonomous AI agents. Every agent action produces a cryptographic bilateral receipt — one Ed25519 signature before execution, one after — hash-chained for tamper-evidence. Those receipts accumulate into Trust Capital: a permissioned machine reputation asset that determines what each agent is allowed to do.
What is Trust Capital?
Trust Capital is a permissioned machine reputation asset built from verified behavior, not vendor claims. Agents earn Trust Capital by staying within their declared covenant, completing tasks reliably, and producing cryptographically verifiable evidence. Higher Trust Capital unlocks higher autonomy, larger transaction limits, lower insurance premiums, and enterprise approval.
What is a covenant?
A covenant is a signed behavioral specification: what an agent may and may not do. It can only be narrowed over time, never loosened, so permissions cannot silently expand.
Does Nobulex work with MCP and LangChain?
Yes. Nobulex is designed alongside the surfaces agents already use—including MCP for tools and integrations such as LangChain. See the integrations section and docs for wiring patterns.
Is Nobulex open source?
Nobulex is MIT licensed and developed in the open on GitHub. You can self-host, audit the protocol, and run the live demo locally.
How does pre-execution enforcement work?
Middleware intercepts actions before they execute, evaluates them against the covenant, and only allows actions that pass. Blocked attempts can still be reflected in the action log for a verifiable trace without executing forbidden behavior.