GitEquity MCP Server
MCP server that lets AI assistants read portfolios and propose trades on GitEquity, with all writes requiring a GitHub-authorized confirmation code for safety.
README
<div align="center">
<img src="./.github/assets/banner.png" alt="GitEquity" width="100%" />
GitEquity
Own real stocks from a GitHub comment.
No brokerage account. No wallet app. No gas.
Website · Docs · Dashboard · X
</div>
The one-line version
You type this in a GitHub issue:
@gitequitybot buy $NVDA 100
A vault gets deployed on-chain under your GitHub ID, 100 USDG gets swapped for gitNVDA through Uniswap V3, and the position sits in your vault until you sell or withdraw. You never opened a wallet. You never paid gas.
Contents
- Why this exists
- How a command becomes a trade
- Security model
- Contracts
- Assets
- Bot commands
- MCP: trade from your AI client
- Repository layout
- Running locally
- Environment
- Related repositories
Why this exists
Every developer already has an identity that is stable, public, and hard to fake casually: their GitHub account. GitEquity treats that account as the primary key for a brokerage.
The consequences are practical rather than philosophical:
- Onboarding is zero steps. If you can comment on an issue, you have an account.
- Identity is permanent. Vaults bind to the numeric GitHub user ID, which GitHub never reuses and never changes. Rename your handle, keep your money.
- The interface is the workflow. Bounties, payouts, and treasury moves happen where the work already happens.
How a command becomes a trade
GitHub comment
|
v
Webhook ──► Intent parser ──► Relayer ──► GitEquityVault
│
Uniswap V3
│
gitNVDA, gitTSLA, ...
issue_commentwebhook fires, signature verified againstGITHUB_WEBHOOK_SECRET- Command text is parsed into a structured intent: ticker, side, size
- Vault is looked up by GitHub user ID, or deployed on first use
- Relayer signs a short-lived authorization and submits the transaction, paying gas
- Vault routes USDG through Uniswap V3 and holds the resulting stock token
- Bot edits its own comment in place with the transaction hash and new balance
Security model
The vault requires two independent signatures for every state-changing call:
| Signature | Held by | Covers |
|---|---|---|
ownerSig |
Per-user keypair, AES-256-GCM encrypted at rest | Vault address, GitHub user ID, nonce, deadline |
relayerSig |
Relayer signing key, hot | Issued only after a GitHub command is verified |
Neither key alone moves funds. The owner key never pays gas and never appears as msg.sender. Replay is blocked by a monotonic nonce plus a five-minute deadline carried inside every signature.
Swap outputs are constrained to an on-chain whitelist, so a compromised relayer cannot route a swap into an attacker-controlled token.
Contracts
| Contract | Role |
|---|---|
GitEquityVaultFactory |
Deploys exactly one soul-bound vault per GitHub user ID |
GitEquityVault |
Custody and swap execution, dual-signature gated |
GitEquityStockFactory |
Deploys the ERC-20 receipt token for each ticker |
GitEquityStockToken |
Soul-bound receipt: gitNVDA, gitTSLA, gitAAPL, ... |
GitEquityToken |
Native EQUITY token |
GitEquityAirdrop |
Merkle distributor for EQUITY |
Sources and deploy scripts live in gitequity-contracts. Addresses are published to deployments.json after each deploy.
Assets
62 tokenized equities on Robinhood Chain, each verified on-chain through symbol() and decimals(). The registry lives in packages/rwa/src/registry.ts.
A sample of what is tradeable:
| Ticker | Receipt | Ticker | Receipt |
|---|---|---|---|
| NVDA | gitNVDA | TSLA | gitTSLA |
| AAPL | gitAAPL | MSFT | gitMSFT |
| AMZN | gitAMZN | GOOGL | gitGOOGL |
| META | gitMETA | COIN | gitCOIN |
| AMD | gitAMD | AVGO | gitAVGO |
Settlement is USDG. Pools are the 1% fee tier for single names, 0.3% for index products.
Bot commands
Tag @gitequitybot in any issue, pull request, or discussion comment.
| Command | Example | Effect |
|---|---|---|
buy $TICKER AMOUNT |
buy $NVDA 100 |
Buy $100 of NVDA |
sell $TICKER AMOUNT |
sell $TSLA 50 |
Sell $50 of TSLA |
sell all $TICKER |
sell all $AAPL |
Close the whole position |
swap $FROM $TO |
swap $NVDA $TSLA |
Rotate one position into another |
portfolio |
portfolio |
Holdings, cost basis, current value |
balance |
balance |
USDG balance |
deposit |
deposit |
Your deposit address and instructions |
withdraw $TICKER |
withdraw $NVDA |
Send tokens to a linked wallet |
confirm <code> |
confirm 8F2A |
Approve a write initiated from an AI client |
Try them without spending anything in the playground.
MCP: trade from your AI client
GitEquity ships an MCP server, so Claude, Cursor, ChatGPT and others can read your portfolio directly.
Reads are direct. Writes always require a confirmation code posted from your GitHub account - the assistant proposes, you approve. An AI client cannot move your money on its own.
{
"mcpServers": {
"gitequity": {
"url": "https://gitequity.xyz/api/mcp",
"transport": "streamable-http"
}
}
}
Per-client setup guides: gitequity.xyz/mcp.
Repository layout
gitequity/
├── apps/
│ ├── api/ Express server, webhook handler, relayer, MCP endpoint
│ ├── web/ React + Vite frontend: landing, docs, dashboard
│ └── social-cards/ OG image generator
├── packages/
│ ├── db/ Drizzle schema and migrations
│ ├── rwa/ Asset registry and price feeds
│ ├── rh-swap/ Uniswap V3 quote and calldata helpers
│ ├── mcp/ MCP tool definitions and handlers
│ ├── x402/ x402 payment protocol
│ ├── api-spec/ OpenAPI spec, single source of truth
│ ├── api-zod/ Zod validators generated from the spec
│ └── api-client-react/ React Query hooks generated from the spec
├── scripts/ Dev and deploy tooling
└── tests/
The API contract flows one way: edit packages/api-spec, regenerate api-zod and api-client-react. Server and client cannot drift.
Running locally
Requirements: Node 20+, pnpm 9+, PostgreSQL 15+
git clone https://github.com/gitequityorg/gitequity.git
cd gitequity
pnpm install
cp .env.example .env # fill in the values below
pnpm --filter @workspace/db run migrate
pnpm dev
API comes up on :3000, web on :5173.
To exercise the bot locally, point your GitHub App webhook at a tunnel such as ngrok http 3000, then comment in a repository where the App is installed.
Environment
| Variable | What it is |
|---|---|
GITHUB_APP_ID |
Numeric ID of your GitHub App |
GITHUB_APP_PEM |
App private key, newlines escaped as \n |
GITHUB_CLIENT_ID |
OAuth client ID |
GITHUB_CLIENT_SECRET |
OAuth client secret |
GITHUB_WEBHOOK_SECRET |
Shared secret for webhook signature checks |
SESSION_SECRET |
Session signing key, 32+ characters |
ENCRYPTION_MASTER_KEY |
32-byte hex, wraps every stored user private key |
DEPLOYER_PRIVATE_KEY |
Deploys vaults and receipt tokens, pays gas |
RELAYER_SIGNING_KEY |
Signs execution authorizations |
DATABASE_URL |
PostgreSQL connection string |
ENCRYPTION_MASTER_KEY cannot be rotated without re-encrypting every stored key. Back it up before going anywhere near production.
Related repositories
| Repository | Contents |
|---|---|
| gitequity-contracts | Solidity sources, Hardhat config, deploy scripts |
| gitequity-mcp | Standalone MCP server |
| gitequity-mobile | Expo app |
| autogit | AI-assisted GitHub App builder |
| playground | Safe repository for trying bot commands |
License
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.