Confidential token launch platform on Starknet

What this enables

• →Launches where the cap table isn't public at t=0. A token launch is the single most attention-dense moment in a project's lifecycle. Today every allocation, every buyer, every insider wallet is legible on-chain within the first block. Confidential participation lands allocations in encrypted notes: the curve fills, price prints, but who bought how much is not attributable.
• →Bonding curves without the whale-attribution cascade. A 200K USDC buy moves the curve. On Pumpfun that buy is a Nansen alert and a copy-trade signal before it confirms. Here the price action is public and the buyer is not - no wallet-level attribution, no reflexive front-running of the launch itself.
• →Dark liquidity for the graduation moment. Projects seed and migrate liquidity without broadcasting the exact size, timing, and provenance of every LP position. The market sees depth; it does not see which sub-account seeded it.
• →Chain-abstracted participation from EVM and Solana. A buyer arrives with USDC on Base or SOL on Solana and participates without bridging, wrapping, or holding a Starknet wallet. Starknet is the privacy-and-settlement layer; the participant never sees it.
• →Insider allocations that can't be reverse-engineered on day one. Team, treasury, and strategic allocations settle into unlinkable execution identities. The public sees supply and price; it cannot map the concentrated allocation to a single dumping wallet the moment the token trades.

What you build

A LaunchFactory deploying per-token confidential bonding curves, each a privacy_invoke helper handling buy / sell / graduate against notes rather than public balances. Participation routes external capital through Chain Abstraction into the Privacy Pool, the curve computes tokens-out inside an Enclave so allocation size never touches a public event, and returns them as encrypted notes. Buyers fan activity across Private Sub-Accounts so no observer links participation across launches or aggregates a profile. At the market-cap threshold, graduate migrates liquidity to an AMM with the curve's final price while the seeding positions stay unattributed. The social feed and price oracle read public curve state - amount, timestamp, price, market cap - with the "who" encrypted end to end.

Why this isn't Pumpfun

Pumpfun is the same business with zero privacy: every buy, sell, and dev wallet is fully legible in real time, which is exactly why launches are gamed - snipers, copy-traders, and dev-dump detectors all operate on public attribution. Pumpfun still clears $25–31M/month in fees even in bear markets, so the demand is proven; the missing layer is the one Starknet provides. STRK20 keeps the visible price action that makes a launch a spectacle and removes only the attribution that makes it a surveillance target.

Why this isn't Fjord Foundry

Fjord Foundry runs LBPs and fair launches on transparent EVM chains where the pool, the participants, and every fill are public by construction. Its whole value is auction mechanics, not confidentiality - the allocation and the buyer set are observable as the sale runs. STRK20 owns the privacy layer of token distribution instead: same fair-launch and curve mechanics, but participation, allocation, and post-launch trading are unlinkable, and buyers reach it chain-abstracted from EVM and Solana without leaving their home chain.