about why this exists, what it does, where it goes
See the cat? See the cradle?
In Kurt Vonnegut's Cat's Cradle a child is shown the string figure and cannot find the cat or the cradle in it. There is nothing there but string, and the shapes the string makes between two pairs of hands. The book's invented religion has a word for the group of people your life is actually tangled with, whether or not you know it: a karass. It has another word for the groups people announce they belong to, the nation, the alumni association, the fan club: a granfalloon. A granfalloon is a label. A karass is a set of relationships.
Nearly everything the web tells you about a person is a granfalloon. A verified tick. A follower count. A bio. A badge you can buy. None of it is made of relationships, and all of it can be minted.
attest takes the other view. A key here has no profile, no badge and no score of its own. It is defined entirely by what it has signed, who has vouched for it, and what it turned out to be right about. Pull those strings away and there is nothing left, which is the point. You cannot buy a shape that is made only of other people's hands.
What it is
A public bag of signed statements about URLs and about keys, made by people and by the agents acting for them, and a service that lets any web page use it without running accounts. It is the bottom of a stack whose top is being able to tell what is more likely to be real.
- You sign up once, with a passkey. Your phone or laptop holds the private key. You also get an AT Protocol account and a handle under attest.monster. AT Protocol leaves sign-in to each hosting server, and the reference server uses a password; here the passkey is the sign-in, and it also authorises every key that signs for you.
- An agent can act for you: it asks with its own key, you approve with your passkey and set what it may write, how much and for how long. That permission is public, so anyone can see whom the agent answers to.
- Any site can put an upvote, a comment box, a vouch or a public statement on a page with one script tag. The site keeps no users, no passwords, no database. Static pages on a CDN work.
- Every statement is signed by your key and stored in your own AT Protocol repository, which relays copy. Anyone can check a statement with your key without asking us.
- Counts are a plain cacheable GET. What a person has attested is a plain GET. The records themselves can be read from their repositories by any AT Protocol tool.
Your identity is the key. You can prove a key is yours by naming it somewhere public, and you can also throw it away. A key that behaves badly can be abandoned, and its owner can start over from zero. Reputation belongs to the key. Mistakes are escapable, at the cost of the years it takes to earn standing again.
What is built today
| piece | how |
|---|---|
| Sign-up and sign-in | A passkey (WebAuthn, ES256) plus an AT Protocol account on our data server: a did:plc and a handle under attest.monster, bound by DNS. |
| Silent signing | On sign-in your passkey signs a 30-day delegation to a device key that lives in the site's own browser storage. Records are then signed on your device with no prompt. Each site gets its own device key. |
| Records | vote, comment, statement, vouch, claim, verification, bookmark, as monster.attest.* records in your repository, each with your inline signature. Subjects: a URL, a DOI, a content hash, or another account. Retracting deletes the record. |
| Index | Follows the repository firehose, verifies every record it counts, drops deleted accounts. /read, /by/<did>, /handle/<name>, /site/<host>. |
| Custody | Revoke any session; add a second passkey from a new device, approved by an existing one; remove one with another's signature. |
| Profiles, vouches, proofs | A page per handle; "I know this person" with its cost shown; proof of control over a website, domain, GitHub user or Bluesky handle, verified by the service. |
| Trust from elsewhere | Prove a Bluesky handle and your Tangled vouches and follows and your Bluesky follows come in as trust edges of different strengths. |
| Agents | An agent asks with its own key; you approve on attest with your passkey, narrowing it if you like. The permission (record kinds as AT Protocol permission strings, a daily limit, an expiry) is public at /agent/<id>. Its records go into your repository marked "via" the agent and count as yours, once. Revoke any time. |
| Private sign-ins | Which sites you sign in to is not published: a sign-in names its site only as a salted hash, and appears in the log only once something signed with it does. |
| Limits | Per-address and per-session rate limits on writes. Reads carry cache headers for CDNs. No cookies. |
Why
Nothing on the web scores the people and outlets that feed us. The founder's phrase for it: "I liken this to the food and drug administration, or to a river; I am being fed pollutants and there isn't even any scoring." PageRank scored emitters once. The platforms took it away and score items, or nothing.
From the builder's side the same gap looks different. A small publication should not have to write its own sign-up, votes and comments, and should not have to hand its readers to a platform that then owns the relationship. Every small service rebuilds the same identity layer, badly, and each rebuild is another silo where reputation cannot travel.
Those are one problem. Signed statements are the anchor. Everything else, scores included, can be rebuilt by anyone from the log. The longer argument, with a fungus in it, is the essay Foundations of Trust.
People and their agents
The internet is changing shape. Trust is moving out of the platforms in the middle and into systems with no center, where people vouch for each other directly and anyone can check the record. And "we" are changing. More of what each of us does online will be done by software acting for us, so a person becomes their own will plus the powers they hand to their agents, and can take back.
That is close to how a company works: each employee has their own badge, a job, limits on what they may sign, and a record of what they did, and the company answers for them. attest gives people the same arrangement with their agents, in public. Each agent has its own key. Your passkey signs what it may do and until when. Everything it writes says so. It earns no trust of its own: a hundred agents vouching add nothing that one does not, because their vouches are yours. In a web full of agents, the useful question is less "is this a person?" than "which accountable person stands behind this?", and that is a question signed records can answer.
What is different about it
We read forty-eight prior efforts before writing a line. Labelers on the AT Protocol, Community Notes, the Ethereum Attestation Service, Nostr reactions, Advogato's trust metric, EigenTrust, SybilGuard, Certificate Transparency, the web of trust, Keybase, the outlet raters. None combines ordinary sign-up with a key the person holds, signed statements about arbitrary URLs and keys, a read endpoint a static page can render, and trust computed by bounded flow with burnable identity. Each of the nearest is missing at least two of those. The flow-based trust math was worked out twenty years ago and shelved because nobody had an adversary worth it.
- Emitters are scored, not items. An article inherits the standing of whoever wrote it and whoever vouched for it.
- The score is never global. When scoring lands, every answer will be "trusted as seen from this root", and the asker picks the root. A million fake keys vouching for each other in a cluster you never touched cannot reach you. There is no symmetric sybil-proof reputation function; there are asymmetric ones, and this is one.
- Nobody can be talked down. Negative statements are shown as evidence and never propagate. A key's standing falls only through its own record: claims that resolved false, vouches that resolved false.
- Public by construction. Statements are public records, so any analytics built on them are public too, to competitors as well. There will be no private dashboards.
- Slow trust. Standing rises at a bounded rate. Nothing irreversible should also be fast.
Other work
Nobody builds this alone, and several of these are worth your time before ours. If you know one we missed, the ways in section says how to tell us.
- badge.blue: a CID-first attestation spec for AT Protocol records, inline signatures or remote proof records, with the repository DID bound into the signed payload so a signature cannot be replayed elsewhere. The closest thing to a shared envelope convention in the atproto world; we are adopting its context binding and low-S normalisation.
- Hypercerts: records of impact work on AT Protocol, in the author's own repo, with evaluations that carry a bounded numeric score, acknowledgements that accept or reject a link, badge awards, and a shared
app.certified.*vocabulary whose signature proofs are badge.blue in production. The lexicons our v2 write-out should map onto. - Trustgraphs, Jake Hartnell's project: each community chooses its starting accounts and rules, members sign weighted, revocable vouches, and a seeded PageRank-style program turns them into scores that anyone can check with a zero-knowledge proof. A working example of the scoring layer that sits above records like ours.
- auth.md, from WorkOS: an open protocol for agents to register for services on behalf of people. Our agent hand-off borrows its code ceremony.
- UCAN: capabilities a key delegates to another key, attenuated at each step and checkable by anyone. The general form of what an agent permission is here.
- socialweb.computer: a pre-alpha compute marketplace on AT Protocol whose access policies read a vouch graph ("only keys my graph vouches for may bid"). A consumer of exactly what this service publishes.
- AT Protocol labels and Ozone: signed, revocable annotations any client can subscribe to; our record envelope follows theirs.
- Advogato's trust metric and Raph Levien's metadata HOWTO: flow from seeds through certifications with a capacity bound. The scoring we intend, written in 2000.
- Ethereum Attestation Service: schema, attester, recipient, reference, revocable, expiry; off-chain mode is a clean attestation model minus the chain.
- Community Notes: bridging consensus across a disagreement axis, open source; needs volume we do not have.
- Nostr reactions with NIP-73 targets and nostr-wot: signed reactions to arbitrary URLs today, plus an oracle for social distance.
- Human Passport, Proof of Humanity, BrightID: proof-of-person by stamps, vouch-plus-deposit, or social graph.
- Keybase, Certificate Transparency, the PGP web of trust, Webmention, Hypothesis, Farcaster's per-app signer keys: each contributed one idea here, named on the technical page.
- The theory: Friedman and Resnick, Cheng and Friedman, EigenTrust, SybilGuard, SybilRank, TrustRank.
The full catalogue, 68 entries with the URL read and the date, is in the repository under reference/.
What comes next
attest is the bottom layer: signed statements by people and their agents. Above it comes scoring that follows trust outward from people you choose; above that, tools that tell you how likely something is to be real, and why. The roadmap has what is built, what is next, and the longer list: scores relative to you on any post, voting, place-based attestations, and the projects we are still reading. attest moved onto AT Protocol a day after launch; the changes page explains why.
Ways in
This project is useful in more than one way, and not all of them involve using it.
- Take the approach. Read the source (MIT), copy the shape, or just the idea: let users sign their own statements with a key they hold. A signature made here matches across every project that checks it, which gives a person durable control over their record that walled gardens do not offer. Enough projects doing this, and the web of trust people have talked about since 1992 starts to exist.
- Critique it. This was built fast, from a reading list. What is missing? Who did it better, and why have we not heard of them? Tell us. We would rather find the effort we missed now than later.
- Attack it. Mint a thousand keys and vouch for each other. Burn an identity and try to come back with standing. Steal a delegation and try to use it from another site. The threat model is on the technical page; break it before the readers arrive.
- Build on it. Put the script tag on a page and stop running accounts, passwords and a comments database. Outsource your sign-in and your attestations, keep your content.
- Use it. Sign up, upvote, vouch for people you know. Every signed statement is one more string in the cradle.
- Read it. The log is public and the score is deliberately not part of the service. Compute your own root's view, study the graph, feed it to a retrieval agent. Competing scorers over the same records are the design.
- Run one. Host an instance, or index the records yourself from the firehose. The promise that identities and statements outlive this service is stronger for every copy someone else holds.
- Fund it. Donations go to
anselm.eth. There is no company behind this, no token, and nothing for sale. - Improve it. Contribute on GitHub, or point us at the similar effort you are already part of. We would rather join than duplicate.
What it will not do
- Publish a single number called "trustworthy". Every score names its root.
- Carry a left-right axis, or any party colours. Neutrality here is arithmetic.
- Sell, log or retain what anyone reads. The planned browser extension is designed so we cannot know.
- Keep private metrics. If it is in the log it is public; if it is not in the log we do not have it.
- Bind a key to a person against their will. You can leave. It will cost you years, and that cost is what makes the standing of everyone who stays mean something.
Lineage
The fuller story, with the people and dates, is on the history page.
Friedman and Resnick, The Social Cost of Cheap Pseudonyms (the newcomer-at-zero result, and the cost we knowingly pay). Cheng and Friedman, Sybilproof reputation mechanisms (no symmetric sybil-proof score exists; asymmetric flow from a source can be). Levien, the Advogato trust metric. Kamvar et al., EigenTrust. Yu et al., SybilGuard. Cao et al., SybilRank. Gyöngyi et al., TrustRank. The AT Protocol label record and NIP-73 target grammar. Community Notes and the per-rater quality weight. Vonnegut, Cat's Cradle, 1963.