MemePortfolio Examples

Python examples for the MemePortfolio MCP Server — build optimal Markowitz portfolios of Solana meme tokens and analyse existing wallet holdings.

Links: Website · Twitter · Blog · MCP Repo

---

MCP Server

URL	https://mcp.memeportfolio.io/mcp
Transport	Streamable HTTP (POST /mcp)
Auth	x402 pay-per-call — no API key required
Network	SOLANA

Tools

Tool	Description
create_markowitz_portfolio	Builds an optimal Markowitz mean-variance portfolio of Solana meme tokens. Input: number_of_tokens (default 10, min 2, max 50).
get_portfolio	Returns balances, prices, USD values, and risk metrics for a Solana wallet's meme token holdings. Input: wallet (Solana base58 public key).

---

Repository Structure

python/
  memeportfolio_http.py   shared helper — MCP connection and agentic loop
  agents/                 agent-based examples (agent .md drives behaviour)
    portfolio_builder.py  create_markowitz_portfolio via agent system prompt
    portfolio_analyzer.py get_portfolio via agent system prompt
  MCP/                    direct MCP examples (plain prompt, all tools)
    portfolio_builder.py  create_markowitz_portfolio
    portfolio_analyzer.py get_portfolio
  scripts/
    run_all_examples.py   run all 4 examples and print a summary table
  requirements.txt

.claude/agents/           Claude Code subagent definitions (from portfolio-mcp)
  memeportfolio-portfolio-builder.md
  memeportfolio-portfolio-analyzer.md

Two example patterns

Agent-based (python/agents/) — the .md agent definition is loaded as the system prompt. Tool selection, output format, and thresholds are defined in the .md, not in code. Use this pattern when you want behaviour that stays in sync with the upstream agent definition.

Direct MCP (python/MCP/) — a plain user prompt with all MCP tools available. Use this pattern for quick integrations or when you want full control over the prompt.

---

Setup

git clone https://github.com/MemePortfolio/examples.git
cd examples
pip install -r python/requirements.txt
cp .env.example .env
# edit .env and add your ANTHROPIC_API_KEY

---

Running Examples

Run all commands from the project root.

# agent-based
python python/agents/portfolio_builder.py           # 10-token portfolio (default)
python python/agents/portfolio_builder.py 5         # 5-token portfolio
python python/agents/portfolio_analyzer.py <wallet>

# direct MCP
python python/MCP/portfolio_builder.py
python python/MCP/portfolio_builder.py 5
python python/MCP/portfolio_analyzer.py <wallet>

# run all examples
python python/scripts/run_all_examples.py

---

Environment Variables

export ANTHROPIC_API_KEY="your-anthropic-key"
# MEMEPORTFOLIO_MCP_URL=https://mcp.memeportfolio.io/mcp  # override for local testing

MemePortfolio MCP does not require an API key — access is x402 pay-per-call.

---

Claude Code Setup

Register the MCP server in Claude Code to use the bundled subagents directly:

claude mcp add --transport http memeportfolio-mcp-server https://mcp.memeportfolio.io/mcp

Then copy the agent definitions into your project:

cp .claude/agents/memeportfolio-*.md your-project/.claude/agents/

---

x402 Payment

The MCP server returns HTTP 402 when payment is required. An x402-compatible client settles the payment automatically and retries. For local development, run the MCP server locally (npm run dev in the portfolio-mcp repo) and set MEMEPORTFOLIO_MCP_URL=http://localhost:PORT/mcp.

---

Further Reading

• Build a Memecoin Portfolio — Beginner's Guide
• Markowitz Portfolio Optimisation for Memecoins
• Best Memecoins to Hold in 2026 — Data-Driven Selection Guide for Solana
• Memecoin Risk Management 2026
• Memecoin Portfolio Rebalancing Guide
• Memecoin Portfolio Tracking Guide