HiveBound is the pre-commitment envelope that wraps every inference call with declared identity, scoped intent, and bonded capital — before the model fires. Nine gates. Ed25519 signed. HAHS receipt. x402 settlement on Base.
Every inference call today is a question with no skin in the game. A prompt arrives, tokens fire, the output floats free. No identity. No accountability. No proof.
No declared identity. No authorized scope. No economic commitment. When an output is disputed — in a compliance audit, a legal proceeding, a regulatory review — there is no artifact. There is a log file. Logs are mutable. Logs do not travel with the output.
Every call carries a pre-commitment envelope. Who asked. What they were authorized to do. What capital they put at stake. Signed before the model fires. HAHS closes the receipt. x402 settles on Base. The artifact is offline-verifiable forever.
HiveBound is not a single feature. It is the assembly of six trust dimensions, each mapped to a layer of the call lifecycle.
Gates 1–6 are the base SHOD stack. Gates 7–9 are HiveBound-specific pre-commitment checks. Ordered cheapest first. Stops at first failure.
HiveBoundAdapter wraps any OpenAI-compatible endpoint. Your provider does not change. Your calls become contracts.
import { HiveBoundAdapter, buildPassport } from '@hive/hivebound'; // Build agent passport from Ed25519 private key const passport = buildPassport(PRIVATE_KEY, { agent_label: 'legal-agent-01', tenant_id: 'acme-corp', scopes: ['inference', 'extraction'], }); // Wrap any OpenAI-compatible provider const adapter = new HiveBoundAdapter({ provider_base_url: 'https://api.fireworks.ai/inference/v1', provider_api_key: FIREWORKS_KEY, privateKeyHex: PRIVATE_KEY, passport, bond_config: { amount_raw: '1000', // 0.001 USDC currency: 'USDC', chain: 'base', chain_id: 8453, treasury_bps: 5, x402_channel: null, }, hivemorph_url: 'https://hivemorph.onrender.com', }); // Every call is now a bonded, signed contract const response = await adapter.chat({ model: 'accounts/fireworks/models/llama-v3p1-70b-instruct', messages: [{ role: 'user', content: transcript }], max_tokens: 1024, }); console.log(response.envelope_id); // signed pre-commitment console.log(response.hahs_receipt_id); // corpus-committed receipt console.log(response.settlement_tx); // x402 on Base
| Scenario | Without HiveBound | With HiveBound |
|---|---|---|
| Compliance audit — "what did the model say?" | Log file. Mutable. No signature. | Ed25519 receipt. Offline-verifiable. Tamper-evident. |
| Disputed output in litigation | Nothing provable. | Signed artifact. Corpus committed. Chain of custody intact. |
| Regulated enterprise sale — HIPAA, SOX, EU-AI | Cannot attest to model provenance. | Full provenance chain. Agent identity. Authorized scope. Capital bonded. |
| Agent acts on tampered corpus | Undetectable. | DR5 catches it. Detection rate 1.000 vs 0.000 across 500 trials. |
| Multi-agent chain — who called what? | No attribution. | Every envelope links forward and backward. Full chain auditable. |
HiveBound is composable with any inference stack. Your calls keep flowing. They just arrive as contracts now.