Guides to launching tokens on Stable
The short answer: point any MCP-capable AI assistant (Claude, Cursor and others) at https://mcp.openfair.app and it can research tokens, build a launch quote, dry-run the transaction and hand you a review link – while your keys stay in your wallet. The final signature always happens on your screen, never on a server.
1. Connect the openfair MCP server
Add the remote server to your MCP client configuration:
{ "mcpServers": { "openfair": { "type": "http", "url": "https://mcp.openfair.app" } } }
In Claude Code it is one command: claude mcp add --transport http openfair https://mcp.openfair.app.
2. What the agent can do
- Research –
search_tokens, get_token, get_trending_tokens, compare_launch_modes: live data from the openfair indexer with canonical explorer links.
- Prepare –
prepare_fair_launch / prepare_instant_launch validate the config server-side and return a full cost quote (creation fee, gas estimate, required value) with a config hash.
- Verify –
simulate_launch dry-runs the exact transaction against the live chain: balance, self-referral and contract checks, with human-readable revert reasons.
- Track –
get_launch_status reports confirmations, indexing and bonding-curve graduation progress.
3. The safety model
The MCP server is read-only by design: it holds no keys, signs nothing, and cannot submit transactions. A launch prepared by an agent becomes a one-time launch session – a link to a review page on openfair.app that shows the frozen parameters (network, factory address, token, fees, config hash). You read them and sign in your own wallet; the parameters cannot change after the quote. Token names and descriptions returned by tools are treated as untrusted data, never as instructions.
4. Why launch through an agent at all?
Because the boring parts – curve math, fee quotes, gas estimation, failure simulation – are exactly what agents are good at. You describe the token; the agent brings back a validated, simulated, costed launch plan; you sign once. Developers can go deeper with the openfair API and SDK.