Build Privacy-Through-Noise tooling for Solana 🌫️🔮

Idea behind the bounty

On a public ledger, everything you do is legible. Every swap, stake, LP, bridge, and transfer is permanent, public, and — increasingly — fed into AI-powered analytics that cluster wallets, attribute identities, and front-run intent in real time. Explorers map your graph. Copy-traders shadow your moves. Surveillance models deanonymize you faster every month with stronger AI systems.

This bounty isn't about hiding funds. It's about behavioral obscurity — making protocols, agents, and users less obvious and far harder to read for automated chain analysis. Privacy through noise. Privacy through plausible deniability. Privacy through longevous crowds.

Superteam Brazil wants to power up open-source privacy tooling for the Solana ecosystem — real, production-grade tools that builders can actually deploy, not proofs of concept.

🦀 This is a Rust-only bounty for hardcore privacy devs. We're looking for serious systems engineers who can build adversary-resistant tooling — not weekend prototypes. If defeating modern chain-analysis is your idea of fun, this is for you.

Who this is for

The tooling we're about to build serves anyone who doesn't want to be an open book on-chain:

• Algotraders who don't want their strategies reverse-engineered

• Whales who don't want every move shadowed and front-run

• Market makers protecting flow and inventory

• Protocols & agents that need to operate without broadcasting their entire playbook

• Everyday users who simply don't want to be clustered, profiled, and tracked

The mission

Build and contribute to three open-source tools that manufacture believable noise, fuzz on-chain intent, and pool behavior into shared anonymity sets. Take them from idealized repos to tools people will genuinely run in the wild.

The three repos

1. account-cooker — believable autonomous activity at scale

A Solana account cooker: spawn agents across the network that stay alive for long periods and behave like real, human users. They wake up at random hours, trade tokens, stake, LP, bridge, flip NFTs, vote in DAOs, interacting with every major Solana protocol the way an actual person would.

The focus is the agent. It should be:

• Easy to manage and strongly scalable (think thousands of agents running for weeks)

• Trivially customizable to integrate new protocols and interaction types

Under the hood, it continuously manufactures believable interaction graphs between wallets:

• Splits funds across many accounts

• Uses realistic, human timing patterns

• Creates recurring but non-deterministic transfers

• Interacts with major protocols

• Periodically consolidates and redistributes assets

• Operates all the way down to dust-level interactions

Goal: make wallet attribution and clustering genuinely hard — so explorers, analytics platforms, and copy-traders can't determine which wallets belong to the same entity. This is a full suite, aimed at real scenarios, not a demo.

2. supersonic-tx — fuzzy transaction bundles

Cast fuzzy tx bundles to mislead voyeurs. Bundle instructions together to throw off stalkers and copy-traders watching your every move — your real intent buried inside plausible decoys so observers can't cleanly separate signal from noise.

Likely shipped as a program — either a single global program everyone routes through, or fresh programs deployed on demand for each use. We want builders to explore the design space:

• Bundling and decoy-generation strategies

• Instruction-level obfuscation that survives mempool and post-hoc analysis

• Making the noise statistically indistinguishable from the real action

• Composability — so other tools (including account-cooker) can cast through it

If you've got a sharper architecture, show us. This one rewards creativity in the design.

3. mirror-pool — crowd-sourced privacy set

A shared anonymity network for behavior. Users voluntarily pool their wallet activity so that actions become collectively deniable.

For example: 1,000 users periodically perform synchronized protocol interactions. Observers can see that an action happened — they just can't easily determine which participant initiated it.

Think "Tornado Cash for behavioral patterns and withdrawals" — not for funds.

The larger the pool and the tighter the synchronization, the stronger everyone's privacy. Build the coordination layer, the privacy set, and the incentives that keep people in it.

What makes a great contribution

• Rust, end to end — all submissions must be written in Rust. On-chain programs, agents, and tooling alike.

• Production-grade and built for real scenarios — robust, tested, deployable, not a toy

• Scalable & customizable — easy to extend with new protocols and interactions

• Realistic — noise that actually defeats modern clustering and copy-trading, not naive randomness

• Well-documented — clear README, install path, and architecture notes

• Open source — MIT, ready for others to build on

Eligibility

• 🇧🇷 Brazilian builders only.

• 🦀 Rust only — all contributions must be in Rust.

• 🥷 For hardcore privacy / systems devs — strong Rust, an adversarial mindset, and comfort with low-level Solana internals. This bounty is deep-end, not beginner-friendly.

How winners are chosen

Three winners — one champion per repo. We'll pick the strongest contributor to each of account-cooker, supersonic-tx, and mirror-pool, judged on:

• Impact — does the work meaningfully advance the tool's mission?

• Quality — production-grade engineering, not slop

• Contribution volume — depth and breadth of what you shipped

Your contributions across any or all of the three repos count. Go deep on one or spread across all three — both paths can win.

Rewards

3 winners — 1300 USDG per repo.

Resources

• Superteam Brazil GitHub: https://github.com/solanabr

• account-cooker, supersonic-tx, mirror-pool (repos — links live on the Superteam Brazil org)

Contact

Reach out to @kauenet with any questions about this listing.