Model Context Protocol (MCP) lets AI assistants call external tools during conversations. Instead of copying data from a browser tab into a chat window, the AI queries APIs directly and reasons over the results in real time. For Solana research, this means you can ask Claude or Cursor questions like "which KOLs are buying aggressively right now" and get a live answer sourced from on-chain data — not from training data that is months or years out of date.
The MadeOnSol MCP server exposes the full v1 API surface as MCP tools: token intelligence, KOL coordination, deployer profiles, scout leaderboards, wallet analysis, and the all-DEX trade firehose. Once installed, every MadeOnSol endpoint becomes a tool your AI assistant can call autonomously.
Installation
Install globally via npm:
npm install -g mcp-server-madeonsol
Or install from Smithery with one click.
Configure in Claude Desktop
Add to your Claude Desktop MCP settings (claude_desktop_config.json):
{
"mcpServers": {
"madeonsol": {
"command": "mcp-server-madeonsol",
"env": {
"MADEONSOL_API_KEY": "msk_your_key_here"
}
}
}
}
Configure in Cursor
Add to your Cursor MCP settings (.cursor/mcp.json in your project or global settings):
{
"mcpServers": {
"madeonsol": {
"command": "mcp-server-madeonsol",
"env": {
"MADEONSOL_API_KEY": "msk_your_key_here"
}
}
}
}
Get an API key at madeonsol.com/pricing. Free tier gives you basic token lookups. PRO and ULTRA unlock KOL data, deployer intelligence, and streaming endpoints.
Workflow 1: Live KOL Activity
Prompt: "Show me which KOLs are buying aggressively right now"
The assistant calls the kol-coordination and kol-tokens-hot tools. The coordination endpoint returns tokens where multiple tracked KOLs entered within a short window. The hot tokens endpoint returns the tokens seeing the most KOL activity ranked by volume and count.
What you get back is a structured list: token name, mint address, number of KOLs who entered, their tiers (S/A/B/C), total volume, and the time window. The AI can then rank these by conviction (S-tier count) or volume, and explain which tokens have the strongest smart-money consensus.
Follow-up: "For the top token, show me the KOL consensus breakdown"
The assistant calls kol-consensus for that specific token and returns the full breakdown — how many KOLs bought vs. sold, average entry price, position sizes, and the consensus direction (bullish, bearish, or mixed).
Workflow 2: Deployer Due Diligence
Prompt: "Analyze this deployer's track record: [wallet address]"
The assistant calls deployer-profile to fetch the deployer's history: total tokens created, graduation rate, how many tokens survived past 24 hours, average peak market cap, and reputation score. Then it calls deployer-trajectory for the trajectory over time — is this deployer improving or declining?
This is particularly useful when evaluating a new token. Instead of manually looking up the deployer on a block explorer and counting successes, the AI does it in seconds and presents a clear verdict.
Follow-up: "List all tokens this deployer launched in the last 30 days"
The assistant calls deployer-tokens and returns each token with its current status: bonded, graduated, peak MC, current MC, and whether KOLs participated. You can quickly spot patterns — does this deployer consistently graduate tokens, or do most die on the bonding curve?
Prompt: "Is this token safe to buy? [mint address]"
The assistant calls multiple tools in sequence:
token-intel — basic token data, market cap, liquidity, holder counttoken-buyers — quality of the buyer base, bot percentage, insider concentrationdeployer-profile for the token's deployer — reputation, graduation ratekol-consensus — what smart money thinks about this token
It then synthesizes all four into a risk assessment. High bot percentage in buyers, a low-reputation deployer, and no KOL interest is a strong avoid signal. A reputable deployer with organic buyers and S-tier KOL accumulation is much more favorable.
No single tool gives the complete picture. The power of MCP is that the AI can call all four and reason across the results in one conversation turn.
Workflow 4: Scout-Based Alpha
Prompt: "Find tokens where S-tier scouts entered in the last hour"
The assistant calls scout-leaderboard to get the current top scouts ranked by performance, then cross-references with recent first-touches data — tokens where high-tier wallets made their first interaction within the last 60 minutes.
Scouts are wallets with a proven track record of entering tokens early and profitably. When multiple S-tier scouts converge on the same token within a short window, that is one of the strongest early signals available.
The AI can filter by scout tier, entry timing, and token age to surface only the freshest opportunities where conviction is high.
What MCP Changes About Research
Without MCP, Solana research involves a loop: open a dashboard, find data, copy it, paste it into a conversation, ask the AI to interpret it, get a follow-up question, go back to the dashboard. Each round-trip takes 30-60 seconds and breaks your flow.
With MCP, the loop collapses to a single conversation. The AI asks for data when it needs it, retrieves it in real time, and reasons over it immediately. A research session that takes 10 minutes manually takes one or two prompts.
This is especially powerful in Cursor during development. If you are building a trading bot and need to understand how a specific endpoint behaves with real data, you can ask Cursor to call it and show you the response shape — no Postman, no browser, no context switching.
Available Tools
The MCP server exposes these tool categories:
- Token tools: lookup, intel, buyers, hot tokens, price history
- KOL tools: coordination, consensus, hot tokens, wallet profiles
- Deployer tools: profile, trajectory, tokens, peak history
