Documentation

The Trustgraphs model, the standard vouching use case, proofs, and governance.

• What is trustgraphs?Trustgraphs is a way to compute useful results from verifiable, relationship-shaped data without asking everyone to trust the computer that did the work.
• How vouch scoring worksThis page describes the standard EAS trust-graph program. It is one concrete Trustgraphs program, not a rule that every trustgraph must use.
• Why trust the resultMost scoring systems ask you to trust whoever runs the computer. Trustgraphs separates the person or service that performs a computation from the evidence that the result is correct.
• GovernanceThe standard vouching wizard always creates a DAO Safe and records it as the network's admin: the onchain authority allowed to change protected rules. The creator's wallet is not the direct admin o…
• QuestionsStart with What is trustgraphs? for the distinction between the platform and its standard EAS vouching program.

The shared proof architecture, network model, and standard vouching mechanics.

• ArchitectureTrustgraphs is proof infrastructure for computations over verifiable graph data. It does not require every network to use Ethereum vouches or PageRank. Each program defines its own source records,…
• Networks and programsA network is one deployed Trustgraphs instance: its input commitments, parameters, authority, proof history, and accepted outputs. A program is the deterministic computation that defines what those…
• Epochs and proofsTrustgraphs outputs advance in rounds called epochs. A checkpoint freezes the input commitments and parameters for one round, a prover runs the program accepted by the verifier active at submission…
• Vouch scoring algorithmThis page specifies the seeded PageRank-style algorithm used by the standard EAS vouch program. It is not the definition of every Trustgraphs program. Contributions, score compositions, and other d…

Create a network, integrate its scores, or operate trustgraphs infrastructure.

• Create a networkThe creation flows deploy independent Trustgraphs instances. Each network selects its initial program verifier, input commitments, parameters, authority, and proof-history contracts.
• Integrate outputsA Trustgraphs network publishes a Merkle root for a program-specific output. Depending on the program, an entry may be an Ethereum-address score, a funding allocation, an AT Protocol node score, or…
• Set up the repositoryTrustgraphs uses several toolchains because the repository includes Solidity contracts, Rust scoring and proving code, a TypeScript frontend and indexer, and local infrastructure. The source code i…
• Run locallyThis walkthrough starts a local chain, storage, database, contracts, proof flow, indexer, and frontend. Begin in a local checkout of the Trustgraphs repository.
• Run a proverThe operator keeps network scores current. It discovers registered instances, freezes eligible checkpoints, reconstructs their inputs, requests proofs, publishes score files, and submits valid resu…
• Run an agentAn agent should have its own funded key and the smallest possible authority. Do not give it the principal's key, a Safe owner key, or a general session key. The useful upkeep calls are already perm…
• Deploy the Sepolia testnetThe Railway project runs four resources in us-west2:
• Deploy to SepoliaA public Trustgraphs deployment combines contracts whose program identity must remain stable with indexing, proving, and publication services that must remain available. Treat it as a security-sens…
• Add a programA scoring program defines how committed inputs become a proven output. Add a program only when the input model, score semantics, or output subject differs from the existing programs. To launch anot…
• Trust graphThe standard trust graph turns signed vouches between Ethereum accounts into a score for each account. It is the default program for communities that want reputation to emerge from their members' r…
• Weighted priorA weighted-prior network combines the EAS vouch graph with an explicit account-and-weight allocation. Its separate trust-graph-weighted program uses that allocation as a persistent personalized Pag…
• Signer syncSigner sync is an optional program and Safe module for a standard, onchain-only vouch network. It recomputes canonical trust scores from the checkpointed vouch history, combines them with recent di…
• HypercertsThe Hypercerts program builds a trust graph from authenticated AT Protocol records published by the Hypercerts ecosystem. It is intended for communities whose useful reputation signals already live…
• Nostr workspaceThe Nostr workspace program turns authenticated activity from a member-scoped Buzz/Nostr workspace into verifiable scores. It supports communities that coordinate in a private or restricted workspa…
• ContributionsThe contributions program proves a funding allocation from two authenticated histories: the parent network's onchain vouches and the round's contribution records. It recomputes reputation and the a…
• Off-chain attestationsOff-chain EAS attestations let members sign vouches without sending an Ethereum transaction. A relay stores the signed payload and anchors an immutable commitment that the trust-graph proof can ver…
• Score compositionsA score composition combines the accepted outputs of several trustgraphs networks into a new, verifiable allocation. It is useful when an application wants to draw on multiple communities or method…

Reproduce public EAS results and check program builds, encodings, and deployments.

• Reproduce a public EAS epochThis walkthrough covers the standard trust-graph program with onchain lane-1 EAS inputs. Those records are recoverable from public chain history, so an independent reader can reconstruct the checkp…
• Golden vectorsTrustgraphs has several proving programs, not one algorithm copied everywhere. Golden vectors lock their consensus encodings and expected outputs so that a change cannot silently make the host, gue…
• Addresses and vkeysNever rely on an address or verification key copied from a chat message or an old screenshot. Resolve deployment data from the tracked manifest for the chain you are using, then check it against th…