Starting an organization on caper takes one transaction and one payment. You connect a Radix wallet, pick a name and a cashtag, and submit; the transaction mints the token, opens the treasury, starts the bonding curve, and hands you a founder badge. The only money that leaves your wallet is the registration fee, which is 10 XRD for a six-character cashtag and rises tenfold for each character shorter. There is no allocation to configure, no treasury to pre-fund, and no application to be approved.
How do I start a DAO on Caper?
Five steps, all of them in the browser:
- Install the Radix Wallet and link it to your browser with the Radix Connector extension. Both are free; details below.
- Connect. Tap the wallet icon in the top navbar and sign the challenge. Signing in is off-ledger and costs nothing.
- Choose a cashtag. One to six characters, uppercase letters or digits. This is the ticker your token trades under, and it fixes your registration fee.
- Fill in the name and description on Create a caper, then submit the transaction. You need the registration fee in XRD on the connected account before this step.
- You are live. The token, treasury, and market exist from that block onward, and anyone can buy in against the curve straight away.
Nothing about the organization has to be decided up front. Members arrive by buying the token, spending is decided by proposal afterwards, and the rules that govern the treasury are the same contract rules for every caper.
How much does it cost to start a caper?
One payment: a registration fee in XRD, priced by how short your cashtag is. Shorter tickers are scarcer, so they cost more.
| Cashtag length | Registration fee |
| 6 characters | 10 XRD |
| 5 characters | 100 XRD |
| 4 characters | 1,000 XRD |
| 3 characters | 10,000 XRD |
| 2 characters | 100,000 XRD |
| 1 character | 1,000,000 XRD |
The contract charges ten to the power of seven minus the length, and rejects the transaction if the payment is even slightly off. Cashtags must be uppercase letters A to Z or digits 0 to 9, and two are reserved by the protocol.
Beyond that fee you pay only the Radix network fee for the transaction itself. There is no listing charge, no monthly cost, no minimum treasury, and no revenue share on the way in.
Where the fee goes. Not to a company. Registration fees are banked in the treasury of the $CAPER caper — the first caper the protocol registered, and one whose own members hold tokens, vote, and can exit on the same terms as any other. The protocol funds itself through a caper rather than around one.
What it costs to run a caper is separate, and it is not a per-caper setting: every caper on the platform trades through one shared logic component that holds a single fee schedule for all of them. The proposal fee is paid into your own treasury. The 0.5% trade fee is charged on both legs, but the two legs land in different places – the buy leg into the treasury of the caper being bought, the sell leg into the treasury of the root $XRD caper, where every sell, swap and exit fee on the platform pools. Nothing is split between the two. Raising funds works both legs through, and Trading covers what happens when a swap fires them at once.
Can the fees change after I launch?
No – not by you, not per caper, and as the platform is deployed today, not by anyone. Every lever is a stored field on the one shared logic component each caper trades through, written once when that component is instantiated and never assigned again. The deployed CaperMain exposes the trade paths, the governance paths, exit, withdraw_founder and a handful of address reads – and not one of them is a setter. The component this replaced had one, update_fees, and it did not survive the August 2026 redeployment.
What the schedule holds today:
| Lever | Live value | What it does |
| Trade fee | 0.5% | Charged on both legs – the buy leg into the caper you buy, the sell leg into the root $XRD caper |
| Collateralization peak | 0.075 | Sizes the founder reward, and decays to nothing by 30% of supply sold |
| Proposal fee | 500 XRD | Flat, paid at creation, kept whether the proposal passes or fails; overpayment is returned |
| Proposal fee rate | 10% | The execution fee on the winning action’s XRD value, charged at the trigger – and waived when the ballot returns “treasury pays” |
| Vote fee | 100 XRD | Charged per ranked ballot, into the caper’s own treasury |
| Vote mint | 1 | Soulbound v minted per ranked ballot cast – the only member-facing source there is. Trading mints none; the vote_rate field this row named until 11 September 2026 is gone from the component altogether |
| Voting period | voting_period_in_seconds | How long the ballot stays open. Drill tuning for the Stokenet deployment, not a production figure – read the live value off the component rather than off this page |
| Market window | proposal_window_in_seconds | How long the market has to veto, running from the trigger rather than from creation. Drill tuning as well – it is declared beside the voting period in the same logic package, and a new package retunes both |
| Baseline window | twap_window_in_seconds | The length of the trailing average that sets the price baseline the window is judged against – a logic constant, retuned the same way |
Those are live values rather than documented ones – the three windows are given as the field to read, because they are tuning that moves – and the point of the column is that you can check them instead of trusting them: every one is a readable field on whichever logic component the registry currently points at. Resolve current_main on the registry and read the fields off the component it names – resolve it there rather than copying an address, because a retired registry keeps answering, and keeps naming its own retired logic, so a stale citation looks exactly as healthy as a live one. An address that names nothing at all is quieter still. The Radix Gateway answers state/entity/details for any well-formed address with an empty items list rather than an error, and an explorer renders a page for it at HTTP 200, so an address pointing at no entity on the ledger passes a link check exactly as a live one does; the only test that separates them is whether the Gateway returns an entity. That warning is worth reading twice, because this page has earned it twice. Until 12 September 2026 it cited a registry that the 11 September redeploy had already retired. The full redeploy of 15 September 2026 retired the next one, and the link above kept pointing at it until 16 September. Both retired registries kept answering. The values above were read on 16 September 2026 off component_tdx_2_1cq6jcm9…cq2wd3na, the CaperMain named by the registry the link now points at.
Changing any of them takes a whole new logic component and a move of the registry’s current_main. That is the platform-upgrade path, and it is open to two callers: the protocol admin badge, and a settled $CAPER UPGRADE proposal, which reaches the registry through propose_main_via_governance – gated to the current logic component rather than to any operator key. The path carries no waiting period on this deployment: the registry’s own upgrade_delay_seconds reads 0. That is now fixed for the registry’s life rather than merely unset – the 11 September 2026 redeploy made the delay a constructor argument and deleted set_upgrade_delay, so not even the admin badge can shorten the window in which holders exit under known-good logic. scripts/redeploy.ts passes 48 hours on mainnet and 0 on Stokenet.
One line this schedule used to carry is gone. Settling a proposal is permissionless, and until 11 September 2026 the caper’s own treasury reimbursed whoever called it – up to 5 XRD a call against a 500 XRD per-proposal allowance pinned to the proposal fee, so a proposal could never cost more to settle than the member who raised it had paid in. That subsidy was removed outright, along with lock_settlement_fee, the allowance ledger and every gas parameter on the crank path. It went because the engine draws locked fees in reverse lock order, so the treasury’s lock was consumed before the caller’s and quietly paid the first 5 XRD of any transaction that contained one settle-path call, whatever else that transaction did. Every crank now pays its own network fee.
What do I need before I can take part?
A Radix wallet, and nothing else. Tap the wallet icon in the top navbar – it carries no text label, and its accessible name is “Connect wallet” – then choose an account and sign the connection challenge — caper uses ROLA (Radix Off-Ledger Auth), so signing is free and produces a session, not a transaction.
Some surfaces are gated by an XRD balance check. Holding any XRD on the connected account unlocks them; you never send XRD to caper to authenticate.
The Radix Wallet is a free, self-custodial app. Install it on your phone, then — on desktop — link it to your browser with the Radix Connector extension so dApps like caper can talk to it. The official portal at wallet.radixdlt.com links every download. Once installed and linked, come back and tap the wallet icon.
How do I join a caper without launching one?
Buy its token. Browse the live capers and buy in directly against the bonding curve — no order book, no counterparty, instant settlement, and the price follows supply rather than a quoted spread. Holding the token is what membership is; there is no separate application or allowlist. How pricing and the fee legs work is in Trading.
How does governance work once I hold a token?
You vote, and then the market votes. Holding a caper’s token makes you a member twice over: you can cast a ranked ballot on any proposal, and your trading is a second, continuous signal – holding is consent, and selling is the vote against.
A proposal names exactly one option kind, and there are six: five executive ones – PAYOUT (send treasury funds to a recipient), INVEST (buy into another caper’s curve), DIVEST (sell a held position back into that caper’s curve), UPGRADE (swap the platform’s shared logic, reserved to the $CAPER caper) and METADATA (rewrite this caper’s own mutable metadata) – and one that executes nothing, DEBATE, a position in an argument that can be ranked and can win outright, and ends there. Raising one takes a proof that you hold at least one of the caper’s tokens, plus the 500 XRD legislative fee.
Phase one. The proposal carries between two and five options, one of which is always a “Do nothing”. Members rank all of them – 100 XRD a ballot, which mints them 1 soulbound v – and a permissionless crank folds the ballots into a Borda tally. The leading option passes only if its share of the weight actually cast clears 1.5 / option_count: 75% on a two-option ballot, 30% on a five-option one. There is no turnout quorum; the bar is a share of what was cast.
Phase two. A legislative pass does not execute, and where the winner is a DEBATE position there is no second phase at all: the contract refuses to trigger one, so the tally is the end of it. Where the winner is one of the five executive kinds, the pass makes the proposal triggerable, and the trigger locks the caper’s trailing average token price as a baseline and opens the market window. When that window closes, anyone can resolve it, and the test is one comparison: the proposal passes if the TWAP over the window itself is at or above the baseline. A market that held is consent; a market that fell is a veto on a decision the members had already taken.
Tallying, triggering, resolving and executing are all permissionless, and each caller pays their own network fee – the treasury reimbursement that used to cover the settle path was removed in the redeploy of 11 September 2026. The history of how this arrangement was reached – including the two days in August 2026 when the ballot was removed entirely – is on voting.
How do I leave and get my money back?
Two routes, both open at any time and neither needing anyone's permission. You can sell back to the curve at the price the curve quotes, or exit outright by handing back your tokens and your vote tokens together – the contract requires both – and take your share of the treasury. That share is (t · v) / (V · T): what you hand in, over the caper's whole vote supply and its circulating tokens.
Vote tokens are earned one way: by casting a ranked ballot, which mints exactly one. Buying, selling, migrating and plain transfers mint none. Trading did mint them, at 0.01 per XRD of gross value, until the redeploy of 11 September 2026 removed the trade-side mint – so the exit now opens on your first ballot and not before. A holder who has only ever bought can still sell back to the curve whenever they like, but cannot claim a share of the treasury until they have voted on something. Influence and exit rights cannot drift apart because they are the same quantity, and that quantity is now bounded by participation rather than by capital.
The exit right is built into the contract rather than granted by a vote. No majority has to approve it and no grace-period window has to be open.
Where next
Start with What is a caper for the model in five minutes. If you're new to DAOs entirely, the knowledge base opens with What is a DAO? and the history of how the industry got here. If you're reading this through an AI agent, Querying a caper from an AI agent documents the MCP servers and the plain-HTTP reads. Questions? Try the FAQ.