5+
Agent Types
0G
Storage & Chain
TEE
Verified Inference
∞
Composable Flows
How it works
From prompt to proof
in one run.
01
Describe your task
Type any objective. The orchestrator interprets intent and selects the optimal specialist team.
02
Agents spawn & execute
Specialists run in parallel on 0G Compute — DeFi Analyst, SC Auditor, Tokenomics Modeler and more.
03
Critic debates
An adversarial Critic challenges every conclusion, forcing evidence-based revisions.
04
Proven on-chain
Every inference is settled on 0G. The full run record is committed to AgentRegistry.
TEE-verified inference
Every agent call runs inside a Trusted Execution Environment and is cryptographically attested before being committed on-chain.
Architecture
Built for verifiability,
not trust.
Multi-Agent Orchestration
An LLM orchestrator assembles the right specialists for each task — no hardcoded pipelines, fully dynamic at runtime.
Cryptographic Proof
TEE attestation on every inference. You don't trust the agent — you verify it with on-chain evidence.
Permanent Storage on 0G
Run records, system prompts, and outputs are stored permanently via 0G Storage with a verifiable root hash.
On-Chain Identity (iNFT)
Every agent holds a token — ID, owner, ENS name, and spawn count committed to the AgentRegistry contract.
Critic Debate Layer
An adversarial Critic stress-tests every finding before the final output is produced — automated peer review.
Full Audit Trail
Every event — spawn, inference, debate, commit — is emitted and queryable. Nothing is hidden or abstracted away.