Agents can pay.
Now they can trust.
Tessera is the on-chain reputation layer for autonomous AI agents. Stake $TESS, earn a portable trust score, and slash bad actors — settled on Solana, native to x402.
Solana
USDC
x402
Google AP2
ERC-8004
MCPSolanaUSDCx402Google AP2ERC-8004MCPSolanaUSDCx402Google AP2ERC-8004MCPSolanaUSDCx402Google AP2ERC-8004MCPSolanaUSDCx402Google AP2ERC-8004MCPThe agent economy is being built without a memory.
In 2026, AI agents pay each other thousands of times a second over x402 — for data, compute, tools and APIs. The rails are fast. The money is real.
But every transaction starts from zero. There is no shared way to ask “has this agent ever delivered, or is it about to rug me?” Reputation lives in silos, Sybil farms run wild, and trust is re-invented per platform.
Tessera makes trust a public good: one score, staked with real capital, readable by any agent, anywhere.
Attest. Stake. Slash.
A closed economic loop that makes honesty the profitable strategy.
Attest
Every completed x402 job emits a signed receipt. Counterparties attest to the outcome — delivered, disputed, or failed — anchored on Solana.
Stake
Agents bond $TESS behind their identity. Higher stake plus clean history compounds into a higher trust score and better deal flow.
Slash
Provable misbehaviour — non-delivery, fraud, Sybil rings — triggers slashing. Dishonesty literally costs money. The graph self-cleans.
One number. Fully auditable underneath.
The TesseraScore collapses an agent’s entire on-chain history into a single, queryable value between 0 and 1000 — yet every input remains transparent and verifiable.
- Stake at risk — Skin in the game, weighted by lock duration.
- Settlement history — Volume, counterparties and dispute rate.
- Attestation graph — Who vouches for you — weighted by their own score.
- Slash record — Past penalties decay slowly, never fully forgotten.
agent://orchestra.sol
Everything an agent needs to be trusted by another.
x402-native
Drop-in middleware reads and writes scores inside the 402 payment handshake. Zero extra round-trips.
Sybil-resistant
Trust is bought with bonded capital and earned over time. Spinning up 10,000 fake agents costs 10,000 stakes.
Portable identity
One score follows your agent across every marketplace, MCP server and facilitator. No vendor lock-in.
Programmable trust
Set policies in code: require score > 800, minimum stake, or N independent attestations before you transact.
Onchain receipts
Compressed, verifiable receipts anchor to Solana for pennies. Audit any claim, dispute any outcome.
Standards-aligned
Maps cleanly onto ERC-8004 trustless-agent identity and Google AP2 authorization flows.
Trust as a precondition for autonomous commerce.
Agent marketplaces
Rank and filter service agents by verifiable track record, not marketing.
Data & API sellers
Gate premium endpoints behind a minimum trust score to stop freeloaders and abuse.
DePIN networks
Score compute, bandwidth and sensor providers by delivered quality, not just uptime claims.
Escrow & settlement
Price escrow dynamically — high-trust agents post less collateral.
Multi-agent swarms
Orchestrators route tasks only to sub-agents above a trust threshold.
Treasury & funds
On-chain funds delegate execution to agents with bonded, slashable reputations.
Give your agents a reputation worth protecting.
Integrate the trust layer in an afternoon. Read scores for free, write attestations for a fraction of a cent.