Skip to content

Update proposal with milestones & payouts, monero-mcp

Baltsar requested to merge Baltsar/ccs-proposals:monero-mcp-q2-proposal into master

What

maintenance, guides, demos, and testing for monero-mcp — the first open-source MCP (Model Context Protocol) server for Monero, giving AI agents a private wallet.

The repo is live and functional (v0.2.0):

  • 12 tools for AI agents (balance checks, transfers, transaction history, address generation, etc.)
  • 5 security layers against prompt injection (allowlist, two-step confirmation, rate limiting, audit log, input sanitization)
  • Available on npm and Docker

A working mainnet demo with Agent Zero has been publicly shown.

This proposal funds:

  • Polishing documentation
  • Creating guides and demos to lower adoption barriers
  • Expanding testing based on real usage
  • Ongoing maintenance

The project remains fully open-source (MIT), free for everyone to use, self-host, fork, or extend — no commercialization or restrictions.

Who

I am Baltsar (GitHub: Baltsar), known in the Monero community for:

  • Designer for Monerujo in the early Android wallet days — UX and visual design, not code. Verifiable via rehrar and anhdres (confirmed in this thread).
  • Designer for Mastering Monero — handled layout, visuals, and user-facing explanations for the community's primary reference book.

I built monero-mcp independently to bring private Monero usage to AI agent frameworks.

Note: on my github opensverige is a chaotic collaboration playground where Swedish AI agent builders experiment and fail fast. It is not a production project and not part of this CCS proposal.

Why

MCP (Model Context Protocol) has become the standard way AI agents interact with external tools. Every major chain is racing to add MCP servers: Solana agents trade autonomously, EVM chains have wallet and DeFi integrations, Bitcoin and Lightning have MCP servers, and market-data MCPs exist for CoinGecko and others. AI agents are rapidly gaining real financial capabilities.

The problem: All existing crypto MCP implementations run on transparent blockchains. Every balance, transfer, and transaction history is permanently visible. For autonomous agents handling payments, data purchases, or decentralized services, this creates permanent surveillance and profiling risks.

monero-mcp is the first and, as of writing, only known open-source MCP server that gives AI agents genuine financial privacy using Monero. Recent developments like native Monero support for HTTP 402 micropayments (e.g., xmr402). Recent work on native Monero support for HTTP 402 micropayments (e.g., xmr402) is actively being built, but not yet deployed at scale. monero-mcp aims to be the natural wallet layer in this emerging stack. This builds on prior experiments such as hyahatiph-labs/tpat and concepts first explored via bounties.monero.social. monero-mcp serves as the natural wallet layer in this stack: agents need tools to hold, spend, and manage XMR autonomously.

Adoption context: AI agents are exploding in 2026 — Gartner forecasts 40% of enterprise applications will embed task-specific agents by year-end (up from <5% in 2025). The global AI agents market is projected ~$10–11B in 2026, with CAGRs of 45–50% through 2030–2033. As agent frameworks (Agent Zero, OpenClaw, Hermes Agent, etc.) gain traction, monero-mcp positions Monero to capture a share of potential new private wallets and transactions in this agent economy — where privacy must remain default.

Early traction confirms demand: monero-mcp is listed on Monerica.com as the world's first MCP server for Monero, used in Agent Zero Docker setups, and highlighted in community discussions.

Effort & Pricing

Estimated ~46 hours of part-time skilled work over 4 months (~3 hours/week).
Rate: ~$135/hr equivalent.
Reference: MA10 XMR price ~$345 at time of writing → 18 XMR ≈ $6,200 USD.

Timeline

4 months (April–July 2026). Work begins after merge to Funding Required; payouts after each milestone's public update (progress report in this MR with links/evidence, followed by community/Core Team approval).

Milestone 1 — Guides & Demos Polish (7 XMR, April 2026)

Deliverables:

  • 5 written tutorials (step-by-step, copy-paste ready): Claude Desktop setup, Cursor integration, Agent Zero Docker setup, OpenClaw integration example, production security hardening guide — lowers the barrier from "Monero developer" to "anyone with a terminal"
  • 2 polished video demos (5–10 min each) demonstrating setup and live usage with Agent Zero and OpenClaw — shows the Monero community and the broader AI/crypto space that private agent payments actually work today
  • Config examples covering Agent Zero, OpenClaw (NanoClaw), Hermes Agent, and Claude Desktop — meet developers where they already are, in the frameworks they already use
  • Updated README reflecting v0.2.0 status, npx install, Docker usage, and security checklist — first thing anyone sees; needs to convert visitors into users

Evidence: Updated repo docs/examples merged to main + CCS update with direct links.

Milestone 2 — Testing & Hardening (6 XMR, May–June 2026)

Deliverables:

  • 20+ documented live mainnet examples with small test amounts (≤0.001 XMR per transfer): balance checks, transfers, history views, edge cases, error handling — proof that it works on real money, not just stagenet demos
  • LLM compatibility matrix documenting how 5+ models (Claude, GPT-4, DeepSeek, Gemini, Llama) handle Monero MCP tool calls — the MCP server works; the interesting variable is how different models behave with it. This is the most requested missing piece from early testers
  • Bug fixes and improvements based on real usage (enhanced audit logging, clearer error messages) — driven by actual mainnet testing, not hypothetical edge cases

Evidence: New docs/examples folder merged + code updates merged + CCS progress post.

Milestone 3 — New Tools & Ecosystem (5 XMR, July 2026)

Deliverables:

  • Add get_tx_proof and check_tx_proof tools to monero-mcp (Monero's cryptographic transaction proof primitives — the building blocks for payment verification and xmr402-style protocols) — these are standalone useful for any payment verification, and the exact primitives needed for the emerging HTTP 402 Monero payment protocol
  • Submit monero-mcp to 2+ MCP server registries/awesome-lists — monero-mcp is currently the only privacy coin in the entire MCP ecosystem; getting listed makes Monero visible to every agent developer browsing these directories
  • Written integration guide: "How to connect monero-mcp to xmr402 payment flows" (positioning Monero in the agent economy) — documents the wallet + payment protocol stack so the next developer doesn't start from zero
  • Respond to community issues/PRs throughout the period — keep the project alive and responsive as adoption grows

Evidence: Code merged to main + registry PRs + integration guide published + final report.

Community Approach

AI agents and MCP integrations are still very new and experimental in the Monero ecosystem — this is frontier work with many unknowns. I don't claim to have all the answers or speak for the community.

I'm excited to collaborate openly: incorporate feedback, iterate based on real usage and suggestions, and evolve monero-mcp together with anyone interested (devs, testers, agent builders, privacy advocates). The goal is a robust, community-owned privacy layer for agents — not my personal vision alone. Looking forward to working with everyone to get this right.

Additional Notes

  • Strong emphasis: Only use very small test amounts on mainnet (≤0.001 XMR per transfer). Never risk funds you cannot afford to lose. This is open community infrastructure.
  • Expiration: October 31, 2026. If not fully funded by then, withdraw or adjust. Any over-funding returns to General Fund.
  • All work stays MIT-licensed and public.

Thanks for considering — open to discussion and revisions in comments!

Edited by plowsoff

Merge request reports