Skip to content
[ ARKIV ]
Arkiv Ideathon August 2026

What can YOU [ ARKIV ] ?

Pitch what you'd build on Arkiv — the queryable, tamper-proof Web3 database. Design the idea, not the deployment.

Each track closes when its window ends · last close Aug 31 · 23:59 UTC

prize pool
$4,000
per track
$1,000
tracks
4
submit
an idea
Queryable Time-scoped Tamper-proof Ethereum-aligned

winners

August is over and the ideas are in. Four tracks, three places each — $500, $300 and $200 — picked by the track hosts against the published rubric.

AI & DevTools

Host: Santiago · window 3–11 Aug 2026 · 80 canonical submissions judged

1st AI & DevTools

Tried That

A shared, self-expiring registry of what does NOT work, so coding agents stop burning tokens rediscovering the same dead ends — confidence is simply how many reports are still alive.

Positive knowledge wants permanence. Negative knowledge wants expiry.
2nd AI & DevTools

Capability Canary

An agent capability index where every claim dies unless an independent prober re-proves it, so the product's answer is an absence nobody can suppress.

The main output is a negative — nobody has verified this agent recently — and it is trustworthy only when the hiding party cannot suppress it.
3rd AI & DevTools

MCP Attest

A neutral, expiring evidence layer where independent scanners, publishers and monitors attest to an exact MCP release and tool-schema hash before an agent connects.

Arkiv does not prove a claim is true, but it prevents one operator from becoming the sole author and lifecycle authority.

Other — the open lane

Host: Santiago · open all month, closed 31 Aug 2026 · 89 canonical submissions judged

1st Other (open lane)

Tideland

A shared game world with no server and no studio, where every building is an Arkiv entity you pay rent on — pine lasts a week, stone a season, a monument five years — and whatever the players stop caring about literally sinks off the map, because the decay is not game code someone can patch, it is the database.

On Postgres, upkeep is a cron job and a promise a patch can break; on Arkiv it is physics.
2nd Other (open lane)

Layak

Indonesian statutory crane and forklift fitness certificates written as Arkiv entities whose lifetime IS the certificate's legal validity, so the gate check at a site entrance is one existence query with no date comparison in it to get wrong.

An expired certificate is not a row with a stale date on it; it is a row that no longer exists to be returned.
3rd Other (open lane)

GelGit

Data that lives on the publisher's own machine and surfaces on Arkiv only during the window it is actually needed, with the timetable itself published on-chain so anyone can query when it comes back.

A window needs no close transaction, no cron, no contract call: its expiry IS its closing time.

Marketplaces

Host: Shantelle Awomoyi · window 12–20 Aug 2026

1st Marketplaces

Still Yours

A listing that stays alive only while the seller keeps proving they still have the item — an unrenewed claim simply disappears, so a sold ticket cannot stay listed on four other venues.

In a double-sale dispute, whoever runs the database decides who was at fault, and that operator is a rival.
2nd Marketplaces

Shared Compute Lease Board

A signed lease ledger for ephemeral agent sandboxes where the lease and the reservation are the same object, so orphaned compute dies on schedule instead of leaking budget and two agents cannot silently hold the same box.

Here neglect frees the resource. Same laziness, opposite blast radius.
3rd Marketplaces

Strikethrough

Every price a listing has been offered at is written as an append-only, self-expiring entity signed by the seller, so the struck-through was-price becomes a fact anyone can verify in one query instead of a claim the seller makes about its own past.

The record's author is the party it indicts.

DeFi

Host: Marcos Miranda · window 21–31 Aug 2026

1st DeFi

Chorus

A witness log, not an oracle — solvers and keepers publish signed, short-lived observations of DeFi state too small or fast-moving to justify oracle-grade economic security, and their standing is only as good as their last corroborated report.

Nothing stops one operator writing three rows and calling it redundancy.
2nd DeFi

PARALLAX

PARALLAX lets a DeFi risk analyst see the blast radius of a protocol failure before it cascades, not after.

The moment you need multiple mutually distrusting observers writing to the same dataset without one of them controlling it, you've left what a plain database can honestly promise.
3rd DeFi

Wallet Guardian

Turn your wallet's invisible token approvals into shared security intelligence, see what each permission can do, how much is at risk, and which real exploit it resembles, powered by an exploit-and-reputation graph that lives on Arkiv so any wallet or dApp can read it.

Native expiry = self-cleaning intel — decay without cron jobs or a moderation DB.

also worth reading

Podium-grade ideas that carry no prize — two because their authors already won in another track (one prize per person, across every entry), one because of a host tie-break recorded in the winners sheet.

No prize Other (open lane)

On the Clock

A 90-day vulnerability disclosure embargo as a transferable Arkiv entity whose lifetime IS the deadline, so the notification date, the party responsible right now, and every request for more time become facts instead of claims.

No prize Other (open lane)

Half-Life

An accessibility map where every fact rots on purpose, so a step-free route shows green only while somebody has confirmed it in the last six hours, and the operator cannot delete the reports that say otherwise.

No prize Other (open lane)

Warranti

A public warranty registry where the entity's lifetime IS the coverage period, so a second-hand buyer verifies a phone with one query and no wallet, and the shop that owes the repair no longer holds the record that decides whether it owes it.

the board

Winners first, then the 50 strongest from AI & DevTools and the open Other lane by the judging panel's score — plus 20 from Marketplaces, which its own host judged. Filter by track, or search all 70.

Showing all 70 ideas

AI & DevTools 1st

Tried That

A shared, self-expiring registry of what doesn't work, so coding agents stop burning tokens rediscovering the same dead ends.

AI & DevTools 2nd

Capability Canary

An agent capability index where every claim dies unless an independent prober re-proves it, so "what is missing" becomes the answer.

AI & DevTools 3rd

MCP Attest

A neutral, expiring evidence layer where independent publishers, security scanners, and uptime monitors attest to the safety and health of exact MCP server releases before AI agents connect.

Other (open lane) 1st

Tideland

An open, serverless game world the tide is always taking back: every wall and monument is an Arkiv entity kept alive by Lifetime Extension, materials are lifetimes (pine lasts a week, stone a season), and whatever the players stop caring about literally sinks off the query surface.

Other (open lane) 2nd

Layak

Statutory inspection certificates for lifting equipment written as Arkiv entities whose lifetime IS the certificate's validity period — so an expired certificate is not a row with a stale date on it; it is a row that no longer exists to be returned.

Other (open lane) 3rd

GelGit

Data that lives on your own machine and surfaces on Arkiv only during the window it is actually needed, with the timetable itself published on Arkiv so anyone can see when it comes back.

Marketplaces 1st

Still Yours

A listing that stays alive only while the seller keeps proving they still have the item.

Marketplaces 2nd

Shared Compute Lease Board

A signed lease ledger for ephemeral agent sandboxes where the lease and the reservation are the same object — so orphaned compute dies on schedule instead of leaking budget, and two agents can't silently hold the same box.

Marketplaces 3rd

Strikethrough

Every price a listing has been offered at is written as an append-only, self-expiring Arkiv entity signed by the seller, so the struck-through "was" price becomes a fact anyone can verify in one query instead of a claim the seller makes about its own past.

AI & DevTools Runner-up

Airlock

Ctrl-C is not a permission system. Airlock makes every agent delegation a signed, self-expiring Arkiv entity delete one link and every sub-agent beneath it loses authority instantly.

Other (open lane) Runner-up

Countersign

Countersign gives two companies integrating over webhooks an independent, tamper-proof record of exactly when something was sent and when it was actually received, so a "we never got that" dispute has a real answer instead of one side's word against the other's.

Other (open lane) No prize

On the Clock

A vulnerability embargo as a transferable entity whose lifetime is the deadline, so the notification date, the party responsible right now, and every request for more time become verifiable facts.

Other (open lane) No prize

Half-Life

A map of accessibility facts that expire on their own, so a step-free route you can see is one somebody actually confirmed today — and an operator cannot delete the reports that say otherwise.

Other (open lane) No prize

Warranti — signed, self-expiring product warranties that outlive the shop that issued them

A public warranty registry: merchants issue coverage as Arkiv entities whose lifetime IS the warranty period — buyers and second-hand buyers verify with one query and no wallet, and the shop that owes you a repair no longer owns the database that says whether it owes you.

Other (open lane) #12

Docket

Funded work with deadlines, recorded so a missed milestone is a signed public event instead of a quiet renegotiation.

Other (open lane) #13

strike

A public calibration record for prediction markets: what a venue priced an event at before it resolved, written where nobody can edit it afterwards.

AI & DevTools #14

Agent Tool Contract Compatibility Board

A shared, self-expiring registry of "framework X actually worked against tool Y version Z last Tuesday" — where compatibility is a signed, decaying claim rather than a green badge somebody stopped maintaining in 2024.

Other (open lane) #15

Lapse

A credential registry with no revocation list — validity is not a field, it is whether the entity still exists, so an expired or revoked attestation cannot be presented as valid because it is no longer there to present.

Other (open lane) #16

Provenance

Provenance lets a smallholder farm land (e.g, cocoa plot) carry its EUDR verification from one buyer/exporter to another. An accredited surveyor verifies the plot once, and other exporters can check that record instead of paying to map the same farm again.

Other (open lane) #17

Trust Window

A live trust graph for real-world safety claims — an inspection goes invalid the moment whoever vouched for it loses their own standing, not just when the paperwork's timer runs out.

AI & DevTools #18

Cross-Client Conformance Ledger

Signed, self-expiring attestations that implementation X conformed to spec revision S — so "which clients are actually ready for this fork?" is a query, not a spreadsheet somebody forgot to update.

AI & DevTools #19

Arkiv Forge

A VS Code and Cursor extension that turns an Arkiv entity model into safe types, valid queries, meaningful lifetimes, and explicit privacy boundaries before the first write

Other (open lane) #20

Hasp

Hasp is a live claim board for heavy equipment serials that shows who currently holds the machine, who published the claim, and when that claim expires.

Other (open lane) #21

Tinkazo — provably fair raffles for communities

A raffle platform for tech communities: attendees import in one click from Luma, the entry list freezes as tamper-proof Arkiv entities, and winners derive from a public randomness beacon — a stadium llama race on top, provable fairness underneath, no wallets needed for participants.

AI & DevTools #22

Assumption Lease

A shared validity layer for AI agents that turns operational assumptions into scope-bound, attributable leases that expire back to “unknown” when they should be rechecked.

AI & DevTools #23

Receipts - Agent Audit Log

A tamper-evident, queryable record of every action your agents take in production - and the permission that authorized it.

Other (open lane) #24

Deadman

Two coding agents on one repo refactor the same file and open competing PRs because neither sees what the other is doing; Deadman gives them a shared ledger on Arkiv where each claim is a self-renewing lease, so live work is visible and abandoned work expires by itself.

AI & DevTools #25

BlackBox — the flight recorder for autonomous agents

A tamper-proof flight recorder for AI agents that act on real businesses: every agent action is signed and written to Arkiv next to an expiring, queryable permission slip — so the party being audited never owns the audit log.

Other (open lane) #26

CivicLibro

CivicLibro is a public, tamper-proof accountability trail for public infrastructure: citizens report problems, authorities log inspections and responses, and neither side can quietly bury the record.

AI & DevTools #27

Eval

Eval prevents coding-agent benchmarks from being rewritten after submissions: organizers commit the hidden exam upfront, while independent runners publish reproducible scores and disputes on Arkiv.

AI & DevTools #28

Agent Leasebook

Principal-signed, short-lived capability mandates that agents, approvers, and MCP/A2A gateways verify before executing side-effecting work.

Other (open lane) #29

Bounty Reality Ledger

A verifiable, self-expiring reality layer for open-source rewards, so contributors can distinguish live bounties from stale, claimed, or settled listings before doing the work.

Other (open lane) #30

ScopeTrace

ScopeTrace is a queryable registry of signed, expiring AI-agent mandates and execution receipts that lets wallets and payment resolvers verify authorization before execution and audit it afterward

AI & DevTools #31

Witness

An AI agent's logs live in a database its operator can edit — Witness is the copy they can't.

AI & DevTools #32

Agent Action Warrants

Agent Action Warrants gives AI agents short-lived, human-signed permission records they can check before sensitive actions, while credentials and execution stay outside Arkiv.

Other (open lane) #33

ContentLease

A creator-controlled permission layer where anyone can verify who may use a creative work, for what purpose, and until when.

Other (open lane) #34

ProofSignal

ProofSignal is a shared evidence layer for crypto opportunities. Projects, researchers, and AI agents can publish time-scoped assessments, attach sources, challenge each other, and let different apps decide what to trust.

Other (open lane) #35

ThreatWindow

A shared, expiring threat-intelligence layer that lets wallets and agents query what trusted security providers knew about an address or domain before a user signed.

AI & DevTools #36

tare

An eval index where a failed run is a recorded state instead of a missing row, so a score can still be defended a month after it was published.

Other (open lane) #37

AuditClock

Smart-contract audit reports as queryable Arkiv entities with a validity clock: the 'audited' badge expires when the audit expires, and nobody can quietly delete or backdate a report.

AI & DevTools #38

Deprecation Clock

Signed, self-expiring support leases that let a coding agent check whether an SDK symbol is still promised to work before it writes the call — so an abandoned API stops being generated even when nobody remembers to deprecate it.

AI & DevTools #39

Crucible

A tamper-proof exercise ledger where red and blue team agents log adversarial security tests as signed Arkiv entities, so MTTD/MTTR can't be edited by either side after the fact.

AI & DevTools #40

Arkitect – The Architecture Compiler for Arkiv

AI can generate architectures. Arkitect proves they're valid.

Other (open lane) #41

CausalGraph

CausalGraph preserves the chain between construction events, evidence and their consequences while a project is happening, so disputes don’t have to reconstruct causality months later.

AI & DevTools #42

CacheFix

A public, self-expiring debugging index in which coding agents look up past verified fixes for the problem they just hit, before spending tokens re-discovering it.

AI & DevTools #43

prove

Prove turns duplicate bug discoveries from autonomous coding agents into independent reproduction evidence before they become maintainer spam.

Other (open lane) #44

Linguator

A community dictionary for endangered Nigerian languages where unconfirmed words expire on their own and confirmed ones survive — so the archive prunes itself without a moderator.

AI & DevTools #45

RedLine

An AI agent that audits your code against real law — GDPR, CCPA, the EU AI Act — quoting the statute verbatim anchored to a file and line and drafting the fix as a PR, while writing its findings and its own self-grades to Arkiv so a reviewer can trust the audit trail without trusting the vendor.

Other (open lane) #46

Cold Chain Custody Trail

A tamper-proof, queryable custody trail for temperature-sensitive goods where every handler signs their handoff to Arkiv and breach evidence outlives the product's shelf life for dispute resolution.

Other (open lane) #47

Passport.ID - Global Digital Product Passport

The open, tamper-proof product passport for circular economy - EU DPP compliant, built for 2027 regulation

AI & DevTools #48

ArcContext

ArcContext is a verifiable company memory layer on Arkiv: AI agents get provenance-checked, time-scoped context from Slack, Notion and email — and auditors or counterparties verify the same memory without trusting a vendor.

Other (open lane) #49

Clearance

Clearance shows whether a worker’s required credentials stay valid for the entire duration of an assignment, not just whether they’re valid today.

Other (open lane) #50

Arkiv Studio

An offline workbench where you design an Arkiv schema and run your real queries against a local mock of Arkiv's engine — catching the model's footguns (string-vs-numeric, no ORDER BY, per-type expiry) in seconds, with no network.

Marketplaces not ranked here

Verified Agent Hire Board

A hire-and-verify marketplace where agent reputation is built from immutable, independently-countersigned records instead of an editable database row

Marketplaces not ranked here

SignalProof

The signal marketplace where the track record can't lie — analysts commit calls on Arkiv before the market moves, stale signals expire themselves, and anyone can independently verify every win rate.

Marketplaces not ranked here

CLAW

An Agentic Marketplace Where Trust is the Currency.

Marketplaces not ranked here

Quaerium

Quaerium is an open request-for-quote marketplace where humans and AI agents purchase expiring, verifiable answers from private datasets without acquiring the datasets themselves.

Marketplaces not ranked here

Beba

Beba is a freight marketplace where posting a load and accepting a quote are the same entity changing hands - the moment a shipper accepts, ownership transfers on-chain from shipper to carrier, turning an open listing into a provable booking. No separate bookings table, ever.

Marketplaces not ranked here

Qevun

A reputation system for Nigeria's informal errand economy, built so trust follows the runner, not the platform.

Marketplaces not ranked here

Terms Receipt

The seller's promise, signed and timestamped at checkout, that stays alive exactly as long as the buyer's claim does and then provably closes itself.

Marketplaces not ranked here

Sealed Alpha

A quant-signal marketplace where every seller's track record is assembled from sealed commitments anchored before the outcome was knowable — so buyers never have to trust the platform, the seller, or the screenshot.

Marketplaces not ranked here

Oreus

``` A verifiable, self-expiring dispute-evidence layer any marketplace can plug in. Both sides commit a hash of their evidence up front, reveal it only if a dispute opens, and every write is signed, so nobody can forge, backdate or delete the record. Fair disputes, less churn.

Marketplaces not ranked here

stall

A marketplace for paid agent services where every listing carries delivery receipts written by the buyers, so a buying agent can filter on what was actually delivered instead of what was advertised.

Marketplaces not ranked here

Turkiv

Tukiv is a self-pruning marketplace that rescues reusable construction materials before disposal deadlines through filterable lots, expiring offers, and creator-verifiable inspections.

Marketplaces not ranked here

SkillSwap Slots

A peer-to-peer marketplace where people and agents post short, self-expiring skill/time availability slots that buyers can query and match with verifiable offers — no stale listings, no fake availability.

Marketplaces not ranked here

Dispute Evidence Packet

A shared, tamper-proof evidence log for marketplace disputes. Both sides attach timestamped proof to an order, and neither can fabricate or backdate what they submitted.

Marketplaces not ranked here

onpoint

A fit-aware fashion marketplace where every listing, quote, and stock signal is a queryable, self-expiring, tamper-proof Arkiv entity — so human shoppers and AI agents browse the same verified inventory.

Marketplaces not ranked here

ProofMarket

An open, self-pruning marketplace protocol where AI agents and service providers publish live capabilities, jobs and offers on Arkiv, while completed-work receipts create portable, tamper-proof reputation that any marketplace can query

Marketplaces not ranked here

Bounty Microboard

A bounty marketplace where dead listings self-prune and reviewer verdicts can be challenged and staked, so reviewer reputation is provable instead of platform-asserted.

Marketplaces not ranked here

Disclosure Window

Researchers list a time-bounded exclusive first look; exclusivity ends when the Arkiv entity leaves the query surface.

The ranked part of this board is the judging panel's first read, not the verdict — the prizes were decided by the track hosts after a second, adversarial pass, which is why the podium and this order differ. The Marketplaces cards are a sample and carry no ranking at all: that track's scorecards belong to its host. All 232 ideas across the three tracks are listed in the repo, one line each: SHOWCASE.md .

why arkiv

Arkiv is the Web3 database: CRUD + built-in indexes over Arkiv entities — filter, paginate and count like you would in SQL. Ethereum-aligned, $GLM-powered.

[ Two ways in ]

prompt

"show me Messi's goals against Real Madrid" await arkiv.buildQuery().where(eq("opponent", "Real Madrid")).fetch()

result

2 goals vs Real Madrid:
goal_311 El Clásico · 23' · 1st half · right foot
goal_207 La Liga · 79' · 2nd half · left foot

prompt

"keep every goal until August 20, then let it expire" await arkiv.createEntity({ goal, expiresIn: 2_592_000 })

result

Stored ✓ · 0x2f81…9c04 · expires Aug 20 · block #512,940.

prompt

"prove goal_311 was really recorded" await arkiv.getEntity("goal_311") // creator, block & tx — checkable on-chain

result

goal_311 · created at block #47,206 by 0xa1b2…88f0 · checkable against Ethereum.

prompt

"where do these goals actually live?" await arkiv.getEntity("0x2f81…9c04")

result

Arkiv DB-Chain ⇄ Ethereum:
arkiv DB-Chain · queryable entities · owner = your wallet
ethereum L1 anchor · final verifiability ✓

tracks

Each track closes at the end of its own window — the open Other lane runs all month. Tap a card for its brief — ideas, data model, host, and when it closes.

schedule

Each track closes at the end of its own window — the open Other lane runs all month. Windows run in sequence and each one closes at 23:59 UTC on its last day.

build

One connection: the Ideathon MCP. Your agent gets the tracks, the rules, the exact judging rubric and Arkiv's fundamentals — plus a prompt that pressure-tests your idea before you submit. No wallet, no key, nothing to deploy.

Step 1

Connect the Ideathon MCP

One URL, read-only, no key. Works with MCP clients that support remote servers — Claude Code, Claude.ai, Cursor and more. It carries everything official — tracks and briefs, rules and rubric, Arkiv fundamentals — so your agent designs grounded, not guessing.

Prompt — paste this to your agent

Connect to the Arkiv Ideathon MCP at https://ideathon-mcp.arkiv.network/api/mcp and show me what's inside — list its tools, resources and prompts with one line each, so I know it's connected and exactly what context we have.

https://ideathon-mcp.arkiv.network/api/mcp What's inside →
Step 2

Work the tracks with real context

Ask for a track's brief and idea seeds, search the knowledge base, or pull the scoring rubric — the same 100 points the judges use. Your agent even knows what deliberately stays OFF Arkiv, so it won't over-promise.

Prompt — paste this to your agent

Using the Ideathon MCP, pull the brief for the track that fits my idea and tell me the 3 things the rubric rewards most.

tools: list_tracks · get_doc · search_kb Arkiv docs →
Step 3

Pressure-test with review_my_idea

The built-in idea coach: it places your draft on a readiness band and asks the 3–5 highest-leverage questions on the judges' angles — the Arkiv counterfactual, data & query design, growth, scope. Answer, iterate, then submit sharp.

Prompt — paste this to your agent

Run the Ideathon MCP's review_my_idea prompt on my idea and ask me the questions that make it stronger.

prompt: review_my_idea Then submit it →

New to the entity model? Start with the Arkiv Fundamentals → — it explains exactly what the form asks about. Or connect the MCP above and ask your agent instead: it serves that same doc.

prizes

Pitch the idea and its Arkiv data model — no deploy. Every track has its own $1,000 USDC pool — $4,000 in total. Each track closes at the end of its own window — the open Other lane runs all month.

How to enter
  1. 1

    Pick a track

    Tap a track card for its brief — buildable ideas + what a strong submission looks like. Or bring anything to the open Other lane. The track you pick sets your deadline.

  2. 2

    Shape the idea

    Map it to Arkiv: which entities, which attributes, which queries. That data model IS the pitch.

  3. 3

    Answer the form

    Your form answers ARE the submission — no repo, no deploy. Optional: a mock, diagram or short video.

  4. 4

    Send it

    Submit before your track's window closes — its host reviews every idea.

Total pool
$4,000

Every track has its own $1,000 USDC pool: $500 for 1st place, $300 for 2nd, $200 for 3rd — $4,000 USDC in total, ranked by the judging team against the published rubric. If a track doesn't receive three ideas that meet the judging criteria, the remaining prize money from that track may be redistributed to other tracks. Strong ideas can also earn mainnet credits, follow-on build support and marketing amplification.

Prizes are paid in USDC to an EVM wallet. Prize winners complete KYC before payout. Winners & highlights roll out the week of Sept 1.

Judged on
Track alignment · 20 Arkiv fit · 20 Data & query design · 20 Impact & usefulness · 15 Clarity & feasibility · 15 Uniqueness · 10
1
Challenge 1 · submissions Aug 3 – 11

AI & DevTools

Give your agent a real data layer. Ephemeral coordination state your agent writes and queries across runs — task queues, review traces, eval indexes, build recipes.

Agents and dev tools constantly need shared state that outlives a single run but shouldn't live forever: what's been claimed, what a review found, how an eval scored. Arkiv gives that state a queryable, time-scoped home with verifiable authorship — no side database, no indexer.

some inspiration
  • 01

    Agent Handoff Queue

    A task queue where agents create, claim, update and close work items as Arkiv entities.

    Why Arkiv · Query by repo / status / priority / expiresAt; stale claims expire; creator metadata verifies who made each transition.

  • 02

    PR Review Trace

    A GitHub Action that stores an agent's findings, fixes and final status per commit.

    Why Arkiv · Query by repo / sha / severity; keep traces only through the review window; tx hashes make review state auditable.

  • 03

    Prompt Run Index

    A tiny eval harness that records prompt inputs, model metadata, scores and failures.

    Why Arkiv · Indexed numeric scores + timestamps make regressions queryable; a short expiry keeps eval clutter from piling up.

  • 04

    Build Recipe Registry

    A searchable registry of build recipes your agent pulls on demand.

    Why Arkiv · Recipes are entities indexed by stack / task / version; maintainers update them and readers verify the publisher.

  • 05

    Incident Context Board

    A shared incident scratchpad for logs, hypotheses, owners and mitigations.

    Why Arkiv · Time-scoped entities fit the incident window; query by service/severity; signed updates keep the timeline verifiable.

What a strong submission looks like

A crisp write-up: the problem, the Arkiv entity + attribute schema you'd use, the queries your agent relies on, and why Arkiv beats a plain database. A mock, diagram or pseudocode makes it sing.

Santiago Zuluaga
Host · Challenge 1
Submit your idea Submissions close Aug 11 · 23:59 UTC
2
Challenge 2 · submissions Aug 12 – 20

Marketplaces

Listings, quotes, bookings, bounties — anything with a lifecycle. Filterable by price/category/expiry, self-pruning when it goes stale, with verifiable authorship on every write.

Marketplaces are mostly state management: what's listed, what's been quoted, what's still available, who said what. Arkiv models each of those as a queryable entity with an expiry and a verifiable creator — so inventory filters cleanly and expires itself instead of going stale.

some inspiration
  • 01

    RFQ Quote Board

    Buyers post requests; sellers post quotes; the UI filters active quotes by price, category and expiry.

    Why Arkiv · Quotes expire automatically; numeric attributes power price/deadline queries; signer metadata verifies quote origin.

  • 02

    Dispute Evidence Packet

    A dispute page where buyers and sellers attach time-bounded evidence entities.

    Why Arkiv · Query all evidence by orderId and side; keep it only through the dispute window; every attachment has a verifiable creator.

  • 03

    Availability Slots

    A booking marketplace for desks, devices or services with expiring availability slots.

    Why Arkiv · Slots are CRUD entities indexed by location / start / price; expired slots prune themselves instead of rotting as inventory.

  • 04

    Bounty Microboard

    A lightweight task marketplace with open bounties, submissions, reviews and status.

    Why Arkiv · Query by skill / reward / deadline / status; owner fields distinguish sponsor, submitter and reviewer.

  • 05

    Seller Signal Feed

    Sellers publish short-lived inventory, SLA or capacity signals buyers can query before transacting.

    Why Arkiv · Time-scoped signals prevent stale claims; attributes make availability filterable; tx-linked updates are verifiable.

What a strong submission looks like

One marketplace flow, designed end-to-end: the entity schema for a listing/quote, the filters buyers run, how state updates and expires, and where verifiable authorship shows up in the UI.

Shantelle Awomoyi
Host · Challenge 2
Submit your idea Submissions close Aug 20 · 23:59 UTC
3
Challenge 3 · submissions Aug 21 – 31

DeFi

Market evidence, risk snapshots, oracle notes, intent feeds. Keep execution off the hot path and use Arkiv as the queryable, tamper-proof state around it.

DeFi generates mountains of state that isn't the swap itself: bids, risk snapshots, oracle candidates, incentive epochs, treasury intents. Arkiv is a good home for that — queryable and tamper-proof, time-scoped to the window it matters in, and explicitly not on the execution hot path.

some inspiration
  • 01

    Auction Evidence Board

    Solvers publish bid metadata, fills and dispute evidence for an auction window.

    Why Arkiv · Query by auctionId / solver / asset / timestamp; evidence expires after the challenge window; creator metadata verifies source.

  • 02

    Liquidation Watch Snapshot

    A dashboard storing periodic risk-state snapshots for mocked lending positions.

    Why Arkiv · Query by market / collateral / healthFactor; snapshots are time-scoped and verifiable, without being execution-critical.

  • 03

    Oracle Round Notes

    Reporters publish candidate values, source links and challenge status for oracle rounds.

    Why Arkiv · Query by asset / round / status; entries live through the dispute window; signer metadata verifies reporter identity.

  • 04

    Incentive Epoch Explorer

    Pools publish incentive parameters per epoch; users query current and previous epochs.

    Why Arkiv · Numeric epoch/pool attributes make incentives filterable; expired epochs prune; updates carry tx hashes.

  • 05

    Treasury Intent Feed

    A DAO publishes swap/payment intents and competing quotes before execution.

    Why Arkiv · Query active intents by token / size / deadline / status; quotes expire; creators are verifiable before a signer acts.

What a strong submission looks like

A DeFi concept with an explicit off-hot-path scope: what evidence/state you'd store as entities, the queries that read it, and how expiry + verifiability become product features.

Marcos Miranda
Host · Challenge 3
Submit your idea Submissions close Aug 31 · 23:59 UTC
Open lane · all month

Other

Doesn't fit the other three? Bring it here. Registries, RWA, provenance, governance, social graphs, game state — if it truly uses Arkiv's model, it belongs.

Not every good idea is AI, a marketplace or DeFi. This track is the open lane for everything else that maps cleanly to Arkiv: registries and RWA, provenance trails, governance state, social graphs, game or world state. The bar is the same — it has to genuinely need queryable, time-scoped, tamper-proof entities, not just any storage. It is also the official catch-all: found the event after your theme's window closed? Enter here — same rubric, competing for this lane's pool.

some inspiration
  • 01

    On-chain Registry

    A queryable registry of anything — domains, licenses, memberships, certificates.

    Why Arkiv · Entities indexed by owner + attributes; expiry models renewals; creator metadata proves who issued each entity.

  • 02

    Provenance Trail

    Track an asset's history as a chain of time-stamped, verifiable entities.

    Why Arkiv · Query the full trail by asset id; each step is a signed write; expiry can scope custody windows.

  • 03

    Governance Log

    Proposals, votes and rationales stored as queryable entities per cycle.

    Why Arkiv · Filter by proposal / voter / status; verifiable authorship on every vote; epochs prune old cycles.

  • 04

    Social Graph Snippets

    Follows, reactions or badges as lightweight, self-expiring entities.

    Why Arkiv · Query by actor + type; ephemeral signals stay cheap; owners are verifiable.

  • 05

    Game / World State

    Off-chain-fast game events mirrored as queryable, verifiable entities.

    Why Arkiv · Query by session / player; the expiry fits a match; verifiable results without a central server.

What a strong submission looks like

A clear pitch for something outside the three named tracks: the entity model, the queries, and a straight answer to 'why Arkiv and not a normal database?'

Santiago Zuluaga
Host · Open lane — bring anything
Submit your idea Submissions close Aug 31 · 23:59 UTC