cookie-mcp
About
An MCP server that gives AI agents full onchain access to Cookie Chain — trade, launch, LP, stake, and bridge to Solana.
Details
- Author
- cookiechain
- Categories
- Developer Tools
Jump to
Setup
Install cookie-mcp in your MCP client (Claude Desktop, Cursor, Windsurf, and others).
Repository: https://github.com/cookiechain/cookie-mcp
Follow the installation instructions in the repository README, then restart your MCP client.
Safe by default: read-only until you add a key, and every money-moving action is simulated before it is sent.
RequiresNode ≥ 22. There is nothing to install or build —npxfetches the published package on first run. Pick your client below. All three use the same server; the only difference is where the config lives.
The quickest way — one command, available ineveryproject:
This registers the server read-only (no key). SeeEnable tradingto add a wallet.
Scopes—claude mcp addwrites to one of three places; choose with--scope:
Use--scope projectonly when you want the servercommitted into a specific repo— it writes a.mcp.jsonthat teammates must approve on first use. For a general-purpose tool like this,--scope useris the right default.
Edit the config file (create it if missing), then restart Claude Desktop:
- macOS—~/Library/Application Support/Claude/claude_desktop_config.json
- Windows—%APPDATA%\Claude\claude_desktop_config.json
Add theserver blockbelow undermcpServers.
Edit~/.cursor/mcp.json(applies everywhere) or.cursor/mcp.jsonin a project (project wins if both exist), then add theserver block.
Claude Desktop, Cursor, and a Claude Code.mcp.jsonall use the identical shape:
-
Config-file clients (Desktop / Cursor /.mcp.json):put it in theenvblock above.
Claude Code:re-run the add with--env(note: this is saved to~/.claude.json; avoid leaving the raw secret in your shell history):
Your key never leaves your machine, is used only to sign locally, and is redacted from all output. Every money-moving action is simulated before it is sent.
Once it's registered, just talk to your agent naturally:
The agent resolves names to mint addresses withsearch_tokens/search_nfts, then acts on the mint — it never turns a name straight into a trade.
Reads(no key):chain_health,get_pools,get_token_info,search_tokens(resolve a token name/ticker to its mint),get_quote,get_wallet(which key this server signs with, and the RPC it uses — no RPC call, so it works when the chain is down),get_balance,stake_info(bCOOK liquid-staking rate / TVL / APY / fees), launchpad readsget_launchpad_pools/get_launchpad_token/get_launchpad_positions, and NFT readsget_nft_listings,search_nfts(resolve an NFT/collection name to a listed mint),get_nft,get_wallet_nfts,get_nft_offers,get_nft_market_stats, and.cookname readsresolve_domain/get_owned_domains/get_domain_listings.
⚠️Before graduation, holdings are program-tracked curve shares, not SPL tokens— they do not appear inget_balanceandtradecannot route them. Exit withlaunchpad_sell, or claim the real token withclaim_launchpadonce the pool graduates; from then on it trades like any other token.
Because those shares are invisible toget_balance,get_launchpad_positionsis the portfolio view: every launch a wallet has a position in, what it is worth on a live curve, and what is unclaimed (tokens after graduation, a Fair-mode refund, a settlement payout, creator fees or vesting). It reads theUserPositionaccounts straight from the chain in batches, so it costs about one RPC round trip per 100 launches. Passownerfor any wallet, or omit it for your own.
A pre-graduation token also has no DEX pool at all, soget_quote/tradewould just report "no route". They now recognise that case and point at the launchpad tools instead, andget_token_infoadds alaunchpadfield when a mint shows no price or liquidity because it is still on a curve.
⚠️Listing escrows the name.list_domainhands the domain to the marketplace's escrow account in the same instruction, so while it is listed the registry reports the escrow as its owner: the seller cannottransfer_domain,update_domainorset_primary_domainon it, and it stops resolving to a payable address. Those tools say so explicitly rather than reporting a stranger as the owner, and passing a listed name where an address is expected isrefused— the escrow is a program account, so paying it would strand the funds.cancel_domain_listingreverses a listing at any time and refunds its rent. There is no re-price instruction: cancel, then list again.
buy_domainrequiresmaxPriceCookfor the same reasonregister_domaindoes — the instruction carries no price argument, so that cap is the only guard. Without it you get the asking price quoted back and nothing is spent.
Use the COOK / native mintSo11111111111111111111111111111111111111112for COOK. Every tool returns JSON; failures return{ error, hint }— never a stack trace, never your key.
yarn install yarn test # lint + format + typecheck + unit tests + boot smoke yarn mcp # run the server on stdio from source (tsx) yarn build # bundle to dist/mcp/server.js (what gets published)
This project is licensed under the terms of the MIT license. See theLICENSEfile.
This is a web browser that enables your coding agent, such as Claude Code, to visit websites on your behalf and assist you in identifying bugs or creating UI test cases.
Create crafted UI components inspired by the best 21st.dev design engineers.
Bring agent evaluations, observability, and synthetic test set generation directly into your IDE for free with Galileo's new MCP server
An MCP server to help AI assistants to answer questions and generate AccelByte Extend SDK code more effectively .
MCP server for AI Diagram Maker — generate beautiful software engineering diagrams directly inside Cursor, Claude Desktop, Claude Code, or any MCP-compatible AI agent
ALAPI MCP Tools,Call hundreds of API interfaces via MCP
AI-powered SVG animation generator that transforms static files into animated SVG components using the Allyson platform
MCP server that gives AI assistants on-demand access to 1,500+ amCharts docs, ~300 code examples, and 1000+ class API references.
APIMatic MCP Server is used to validate OpenAPI specifications using APIMatic. The server processes OpenAPI files and returns validation summaries by leveraging APIMatic’s API.
One shared context layer for AI agents and humans — live API specs, DB schemas, and versioned contracts across repos so every agent and teammate works from the same source of truth.
Build and deploy full-stack Next.js apps with 98 tools for React, AWS, and MongoDB
Sign in to leave a review
Use Google, GitHub, or an email account so ratings stay tied to real people.
No reviews posted yet.





