Our own programs. Real USDC collateral. Permissionless launches.
We are deploying the Clawd Percolator stack — wrapper, insurance stake, and position-NFT programs — directly to Solana mainnet. The full lifecycle has already been rehearsed end-to-end on devnet with a real filled trade. This page is the exact plan: what it costs, what is proven, and what still gates the launch.
⚠ Not live yet. Nothing has been deployed to mainnet and no real funds move through this page. The terminal remains the Clawd Chain testnet (test cUSD, no real value). Mainnet goes live only after the pre-live checklist below clears — funding, the final fee-authority key, and the program-ID decision are the owner's calls.
Route
A — our own stack
Status
Phase 2 ✅ devnet-proven
Total locked rent
10.155017 SOL
Recommended funding
~15.5 SOL
Collateral
Real USDC
// why route A
The other mainnet program can't launch markets
There is already a Percolator-lineage perps program deployed on Solana mainnet (ESa89R5Es3rJ5mnwGybVRG1GrNt9etP11Z5V2QWD4edv). We simulation-tested launching on it before building anything. It is dead for launches — so we deploy our own. The evidence:
Route B — dead
ESa89R… rejects every market creation
The program was upgraded on 2026-05-12; the new binary's slab guard demands exactly 1,525,720 bytes and returns error 0x4 for every other size.
At exactly 1,525,720 bytes the guard passes — but InitMarket burns the entire ~1.386M compute-unit budget and dies.
Exactly one slab exists on it (SOL/USDC, 96,784 B, created 2026-05-06 under the previous binary). Their own post-upgrade tests failed identically. Nothing has launched in ~6 months.
Only their upgrade authority can fix it. It is not ours to repair.
Route A — approved 2026-11-08
Deploy our clawd-percolator stack
All three programs: wrapper (markets + engine), stake (insurance), NFT (transferable positions).
Protocol fee authority patched to our own key in the mainnet build (the rehearsal placeholder never ships).
LP model: operator-run LP service + open LP vault. Pricing: hybrid Pyth for majors, auth-mark for long-tail markets.
// exact program funding
What the deploy costs — to the lamport
Computed from the actual Phase 2 rehearsal builds at the live chain rent rate of 5,080 lamports/byte (rent-exempt minimum = (account size + 128) × 5,080, plus 0.000833 SOL per program account). Programdata space = ELF size + 45 bytes. These figures supersede the earlier ≈9.15 SOL locked / ≈13.9 SOL peak estimates, which were measured from the older Clawd Chain ELFs — the rehearsal builds run +8.2% / +18.0% / +22.3% larger.
Program
ELF size
Programdata rent
Locked (total)
Deploy-step peak
Wrapper — markets + risk engine
1,479,336 B
7.515906 SOL
7.516739 SOL
15.0326 SOL
Stake — insurance fund
285,680 B
1.452133 SOL
1.452966 SOL
2.9051 SOL
NFT — transferable positions
232,992 B
1.184478 SOL
1.185311 SOL
2.3698 SOL
TOTAL LOCKED
10.155017 SOL
fund ~15.5 SOL
■ Wrapper 74.0%■ Stake 14.3%■ NFT 11.7%Locked rent is recoverable only by closing the programs — treat it as locked.
Why funding is ~15.5 SOL when only 10.16 SOL stays locked: a deploy holds the programdata rent plus a same-size buffer while writing. Buffers are reclaimed between deploys, so the binding constraint is the largest single step — the wrapper at 15.0326 SOL — plus fees. Deploy order: wrapper → stake → NFT → wiring.
// phase 2 — devnet rehearsal, complete
Already proven on devnet
On 2026-11-08 all three programs were built, deployed to Solana devnet, and run through the full user-launch lifecycle: launch registration → InitMarket → oracle configure → LP portfolio + LP vault → stake InitPool → second-user portfolio → a real TradeNoCpi that filled (matched basis +1,000,000 / −1,000,000, fee taken) → permissionless cranks. Market mark moved 100 → 90 → 95 under crank.
Exact account sizing: a 1-asset market account is exactly 3,163 bytes (+1,813 B per extra asset); portfolios are 9,555 B. Allocate cannot shrink — the wizard must size first-time, every time.
Sequencing rule: stake InitPool rotates market authority to the stake pool PDA. All market admin (oracle config, fee splits) happens before InitPool.
Oracle: the rehearsal priced via AUTH_MARK because Pyth's devnet update endpoint now needs an API key. Hybrid Pyth wiring is a Phase 3 item, proven before mainnet.
Collateral is a per-market field — the rehearsal used a test mint; mainnet uses real USDC with zero program changes.
// launch economics — once live
What a launch costs a user on mainnet
≈0.02 SOL
Launch a market
Market account rent 0.01672 SOL (3,163 B) + vault ATA 0.00149 SOL + fees ≈ 0.02 SOL, plus your USDC seed for the vault. Rent is recoverable when the market is closed and reclaimed.
0.04911 SOL
Open a trader portfolio
Your own portfolio account rent (9,555 B). One per trader per market; recoverable on close. Your USDC margin sits in the market vault, not in this account.
≈0.00528 SOL
Optional stake-pool set
Insurance stake-pool accounts, only if you run or join a market's insurance pool. Collateral everywhere is real USDC (EPjFWdd5AufqSSqeM2qN1xzybapC8G4wEGGkZwyTDt1v).
// the road to live
Phases
PHASE 0–1 ✅
Recon + verdict
Fork inventoried, engine re-pinned, Route B simulation-tested and ruled dead, Route A approved and scoped to all three programs.
PHASE 2 ✅
Devnet rehearsal
All three programs deployed to devnet; full launch lifecycle with a filled second-user trade. Exact mainnet funding computed from these builds.
PHASE 3 — GATED
Mainnet deploy
Owner funding tap (~15.5 SOL), final fee-authority key, program-ID decision, Pyth hybrid proof, fresh-build sanity rehearsal — then deploy wrapper → stake → NFT.
PHASE 4
Wizard + keeper
Browser-signed launch wizard and registry wired to OUR program IDs, plus the keeper/crank service (cranks need ≥1 live portfolio per market).
PHASE 5
Live
First user launch verified end-to-end on mainnet with real USDC. This page flips from PRE-LIVE to LIVE with the mainnet program IDs.
TODAY
Testnet keeps running
The Clawd Chain terminal stays exactly as it is — free, breakable, no real value — while mainnet is prepared.
// pre-live checklist
What still gates the launch
☐Fund the deployer — ~15.5 SOL. Covers the 15.0326 SOL wrapper-step peak plus fees; 10.155017 SOL stays locked as program rent.
☐Final protocol fee-authority key. Baked into every market at InitMarket. The rehearsal used a throwaway placeholder (G3v2C7pHztZ9XnZbeQhv7Erc5RugEidBJnaF7gXrgB6k) that never ships; the real key is set at deploy and can rotate later via SetProtocolFeeAuthority.
☐Program-ID decision. The Clawd Chain wrapper keypair (6Grr…) is not on this machine — it lives on the Clawd Chain deploy host. Retrieve it, or take the rehearsed patch path with fresh mainnet program IDs. No user-facing downside identified either way; it's an explicit owner call.
☐Unaudited-code gate. These programs have not had an external audit. Going live with real USDC is the owner's explicit call, with the engine diff against upstream (themusebook/percolator) as a pre-mainnet review step.
☐Pyth hybrid wiring proof + keeper service. ConfigureHybridOracle proven against live feeds, and the crank service built (it does not exist yet).
✅Done: Route A approved · all-three-programs scope · engine re-pin 82ca393b · devnet rehearsal with filled trade · exact funding computed · this plan published.