Skills confirmed by the people you worked with and for
Found through friends of friends. A peer-verified contribution network: time contributed to shared projects, signed by the counterparties who received it - no blockchain, no tokens-as-currency, no central authority.
Post a demand - what you need - or a proposal - what you offer. The network pairs demands with proposals across your contacts and hands you a shortlist, ranked by preferences you define on your own device.
A time book and time bank for open-source work · arm64 · sideload, not a store build · the web UI keeps your key on the server - a preview, not the client you trust with real work
How it works
A contribution graph, not a ledger
Every contract runs the same lifecycle - Job Request → Acceptance → Milestones → Completion → Multi-party Sign - and each signed contract is itself a trust relationship. If Alice and Bob completed a job, Alice trusts Bob in that domain; Carol can walk the chain from Alice to evaluate Bob without a central authority computing anything on her behalf.
Post a need, get a shortlist
A demand says what you need; a proposal says what you offer. Post either one, and the network walks your contact graph to pair it with the opposite kind, then returns a ranked shortlist - ordered by the weights and filters you set on your own device, never on a server.
Quant: a unit, not a currency
1 Quant = a quarter hour of work at minimum qualification. It denominates value on a contract but has no ledger of its own - what you hold is your counterparties’ signed records, not a balance.
Web of trust, from completed work
Reputation isn’t a separate score - it emerges from the job lifecycle itself. Trust is domain-specific and computed locally by each participant, with no global credit score.
No blockchain needed
Every contract is bilateral and both parties sign, so omission is provable by production, not by gap analysis. No consensus, no mining, no validator set, no gas token.
Signed records, no consensus
Contribution records are Verifiable Credentials, dual-indexed under both parties’ keys and checked against a publication window - omission is detectable within hours, not assumed absent.
Sybil resistance via cascade block
Trust flows down a signed chain; a block propagates up it. Behind any bot farm sits a limited number of real signing accounts - find and block those, and the farm falls with them.
Neither side has to be online
No step of a contract needs the counterparty - or a network - reachable at the moment you take it; each one is composed and signed from what you already hold. Relays carry the signed records on to the other side whenever their client next wakes.
did:key / did:web + Nostr
Identity is a keypair, not an account with a provider. Storage and transport run over Nostr relays; a client is a thin signing device - the same model Signal uses for messages.
How to join
No gate and no waiting list - install the app, let it generate a keypair, and you are in. Your invite link works anywhere you can post a URL, and someone who follows it arrives as a first-degree contact instead of a stranger several hops away.
Read the joining guideFull design rationale - platform selection, mobile architecture, data model, privacy, and legal framing - lives in abstract.md.
FAQ
Common questions
Is Quant a cryptocurrency?
No. A Quant is a unit of measure - 1 Quant = a quarter hour of work at minimum qualification, analogous to a meter or an hour. It denominates the value of work on a contract but has no ledger of its own; what you hold is your counterparties’ signed records, not a balance or a tradeable token.
Do I need a blockchain to use this?
No. Global ordering is unnecessary - bilateral signatures already make omission provable by production. Consensus is unnecessary - the design accepts detection over prevention. And token transfer is unnecessary, since the unit is time, not a tradeable asset.
Who computes my reputation?
You do, locally, from the public graph of signed contribution records - with your own weights and tolerances. There is no global credit score; there are as many subjective readings of the same public record as there are participants.
How do I find someone to work with?
You post a demand - what you need - or a proposal - what you offer; a demand is matched by someone’s proposal, a proposal by someone’s demand. The query propagates a few hops through your web of trust and each client answers for the people it knows. What comes back is a shortlist, deduped and ranked by the weights and filters you set on your own device; no server decides who you see.
What stops fake accounts or bot farms?
Trust flows down a signed chain, and a block propagates up it. Every account is signed into the web of trust by someone, so behind any bot farm sits a small number of real signing accounts - find and block those, and the whole farm falls with them. Detection is social, not algorithmic.
Does exchanging work through this count as taxable barter?
The structural defense only holds when work is on a declared open-source project and no project involved is controlled by the counterparty in a way that privatizes the benefit - the same shape as ordinary open-source co-authorship. Direct bilateral work-for-work, or contribution to a counterparty-controlled private project, falls outside that framing and is the participants’ own responsibility to assess. This has not been confirmed by a written tax attorney opinion - treat it as an engineering description of the system, not legal advice.
Can I delete something I published?
Deletion here is advisory only. A relay may honor a deletion request, but nothing requires it to, and any relay, contact, or archive that already copied a record may keep it indefinitely. If your jurisdiction grants a legal right to deletion (e.g. GDPR), publishing through this protocol may not by itself satisfy it - don’t publish anything you may later be legally required to delete.
What’s the technical stack?
did:key / did:web for identity, W3C Verifiable Credentials with SD-JWT selective disclosure for attestations, Nostr relays for storage and transport (ATProto as a migration target), and Tauri v2 for Android, iOS, and web from one codebase, with signing delegated to an external signer app.
What’s actually built today?
An early prototype: the protocol layer plus a local referral-routing demo. Nothing here is ready for real transactions or real personal data yet.
More open questions and the reasoning behind each decision are in the full design FAQ.