Trustensor

Trustensor is a Bittensor subnet that provides decentralized, incentive-driven trust scoring for AI agents. As AI agents gain the ability to hold wallets, execute DeFi trades, and hire other agents, the infrastructure for evaluating whether they can be trusted does not exist. ERC-8004's Reputation Registry was built to store trust data -- it is empty. No off-chain provider computes standardized trust scores.

Trustensor fills that gap. Validators send miners an Agent Data Package containing source code, behavior logs, and metadata. Miners compete to build the best trust evaluation models and return a structured TrustScoreSynapse: a trust score from 0 to 100, risk classification, severity level, specific threat flags with evidence, and reasoning. Validators score miners against a curated ground truth dataset of 5,200+ agents drawn from the OpenClaw/ClawHub malicious skills corpus. The best miners earn the most TAO through Yuma Consensus.

Think of it as a dimmer, not a light switch. A DeFi protocol can set their trust threshold at 85. A wallet warns users at 50. A developer sandbox accepts anything above 10. Binary detection flags cannot do this. Trustensor trust scores can.

Revenue comes from a Trust Score API with four tiers (Free to Enterprise). API revenue funds TAO buybacks, creating a dTAO flywheel that aligns commercial growth with miner incentives. 13 Bittensor subnet integrations are already mapped.

Completed during the Ideathon period:

• Full subnet proposal covering incentive mechanism design, anti-cheating measures, ground truth governance, technical specification, business model, competitive analysis, and ecosystem integration map (13 subnets)

• Core subnet scaffold: TrustScoreSynapse and AgentDataPackage schemas (synapse.py), miner forward() template (miner.py), validator evaluation loop (validator.py), and MinerScore scoring function across five dimensions (scoring.py)

• Ground truth dataset design: 5,200+ agent benchmark from OpenClaw/ClawHub malicious skills, synthetic generation pipeline specification, and three-phase governance model

• Competitive analysis confirming no direct competition within Bittensor and a clear gap in the external landscape

Pre-fundraise. Bootstrapped solo founder. Seeking seed funding post-Ideathon to fund mainnet launch, dataset curation, and initial miner recruitment. Target raise: $250,000--$500,000 seed round. Open to strategic investors with presence in the Bittensor ecosystem, AI agent infrastructure, or DeFi security.