
By Rumble Fish Software Development - RFP

An open-source x402 payment facilitator and Bazaar discovery layer for Stellar, letting AI agents find, price, and pay for API endpoints and MCP tools per request in USDC or any SEP-41 asset, at near-zero settlement cost (~0.0023 XLM per transaction). Includes a usage-based ("upto") settlement scheme for metered pricing — proposed upstream to the x402 spec — an MCP-native discovery interface for agent runtimes, and SDK support in both TypeScript and Rust.
$78.0K
Our team operates the Soroban Block Explorer, a live production human-readable transaction and contract explorer for Stellar, delivered under a separate SCF Build Award. We are also currently developing a Prices API (unified price data for classic and Soroban assets) under a second SCF award, on track to ship before this engagement needs to integrate it. Rather than build synthetic demo endpoints for this RFP's required example integrations, we will gate both of these production services behind the x402 facilitator and Bazaar proposed here — giving the infrastructure real, non-trivial production traffic to prove itself against from day one, and directly furthering our own independent monetization plans for both projects. This is concrete evidence of both our SCF delivery track record and our ability to ship production Stellar infrastructure, not a first-time proposal.
[Deliverable 1] Foundation
Facilitator hardened and live on stellar:testnet: verify/settle/supported endpoints, any SEP-41 token accepted, sponsored fees, rate limiting documented. Bazaar resources endpoint and cataloging pipeline live on testnet (ranking not yet enabled). upto scheme design finalized; proxy contract deployed to testnet (pre-audit) with passing unit tests for ceiling/nonce/expiry enforcement. TypeScript SDK helpers and Rust seller crate scaffolding in progress. Core test suite skeleton in place.
How to measure completion: Testnet facilitator responds correctly to verify/settle/supported per the x402 v2 spec for at least one SEP-41 asset; GET /discovery/resources returns a correctly-shaped catalog entry for a seeded test resource; upto proxy contract deployed to testnet with a documented test suite covering ceiling enforcement, expiry, and single-use nonce consumption; a published design note for scheme_upto_stellar.md.
Budget: $15,600
[Deliverable 2] Feature Complete
Bazaar search endpoint live with ranking (relevance + quality score + curation) on testnet. MCP discovery server functional (search + paid-call tools). scheme_upto_stellar.md submitted to the x402 Technical Steering Committee. TypeScript SDK and Rust seller crate both feature complete. Facilitator live on stellar:pubnet. External security audit engaged and underway. Prices API and Block Explorer API both payment-gated on testnet using the new Rust crate.
How to measure completion: GET /discovery/search returns ranked results for a natural language query against at least 3 seeded resources, with documented ranking signals; MCP server responds to a search-then-pay tool call end to end on testnet; TSC submission of scheme_upto_stellar.md is publicly visible (PR or discussion thread); Prices API and Block Explorer API both return 402 responses with valid payment terms on testnet and complete a settled payment end to end using the Rust crate; external audit engagement confirmed and kickoff completed.
Budget: $23,400
[Deliverable 3] Mainnet Launch
Security audit findings resolved. upto scheme merged upstream, or TSC engagement fully documented if merge timing is outside our control. Bazaar and MCP server live on stellar:pubnet. All three example integrations live on mainnet: Prices API and Block Explorer API gated via the Rust crate, plus the canonical Express/TypeScript reference example. Runbook and monitoring live, targeting 99%+ uptime. Conformance report published with settled transaction hashes. Role-based developer guide published to Stellar Developer Docs.
How to measure completion: Facilitator, Bazaar, and MCP server publicly reachable on mainnet at documented URLs; Prices API and Block Explorer API each show at least one real settled mainnet transaction obtained via x402; audit report published with all findings marked resolved or explicitly accepted with rationale; conformance report published listing e2e test results and mainnet transaction hashes; monitoring dashboard accessible; professional user testing completed.
Budget: $31,200
Rumble Fish is a 26-person software house, founded in 2017, that has done blockchain work since its first year. Five of us are dedicated to Stellar work. We have built on Stellar since January 2026, starting with a Soroban smart-contract migration for Merkl that Stellar referred to us. Since then we have received three SCF Build awards, for the Soroban Block Explorer and the Prices API (both live, both open source) and the x402 facilitator, and an SCF Public Goods award. Our SCF work is public at github.com/rumblefishdev.
The engineers who built that work are listed below. This award is sized for about two engineers, who will come from this group depending on timing at kickoff:
Karol Kowalczyk (github.com/karolko9): lead developer of the Soroban Block Explorer, our largest Stellar codebase, and a core contributor to the x402 facilitator.
Oskar Karcz (github.com/karczuRF): lead developer of the Prices API and contributor to the x402 facilitator; built this proposal's evidence repository (corpus entries, generator, the #729 reproduction).
Stan Królikiewicz (github.com/stkrolikiewicz): works across the Block Explorer, the Prices API and the x402 facilitator; author of our internal write-up of the production ingestion failures listed in Traction Evidence.
Adam Kot (github.com/adamkoot): Prices API and x402 facilitator.
Filip Mazur (github.com/Efem67): Soroban Block Explorer.
Marek Kowalski (github.com/kowalski) coordinates our SCF work and is the point of contact.
Both indexers read LedgerCloseMeta straight from Galexie, decode it with rs-stellar-xdr in production, and have had to survive every protocol upgrade since launch, so we have hit the bug classes this RFP targets first-hand (listed in Traction Evidence).

