CLI and agents
The launch CLI is intended to let a creator or agent describe a launch, review the resulting configuration, and authorize the transactions. No Launchpad CLI package, installer, or executable is present in the supplied repositories.
One-command creation
The planned interface includes both a natural-language prompt and a structured configuration file. The following commands are design examples, not runnable commands:
# Proposed: describe the launch and inspect a plan.
launchpad create \
--prompt "Create a bonding curve for ORBIT with 1 million tokens, 40% sale, 30% liquidity, 30% team, USDC quotes, initial price 0.10 USDC, a 30-day team cliff and 300-day vesting" \
--network base \
--dry-run
# Proposed: create from an explicitly reviewed file.
launchpad create --config ./launch.json --network base
There is no installation command yet. Do not install an unrelated package merely because it uses the name launchpad.
From a prompt to an executable plan
The intended CLI translates the request into a structured configuration and validates it before signing. It must surface missing decisions rather than invent a supply, issuer, fee, vesting rule, or payout address. It must also name the network. Base and Stellar are not interchangeable.
A dry run should identify the project and token, issuance controls, distribution account, allocations, quote asset identity, format parameters, dates, vesting, reputation label, and the intended sequence of transactions.
For a curve, a requested initial price must be reconciled with reserve shape, inventory, and the target. For auctions and presales, follow the live Base sale contracts. Stellar sale contracts are part of the expansion.
Agent-friendly behavior
The intended automation interface needs explicit network and signer selection, machine-readable plans and results, a validation-only path, and final transaction receipts. Retrying after an uncertain submission must first reconcile the previous transaction; a repeated command must not accidentally issue the token or purchase twice.
Natural language is an input method. It does not bypass review of token issuance, transfers, fees, or wallet authorization. The exact flags, config schema, package name, and output format remain open.
Available tooling today
The Stellar contract repository includes scripts/deploy.example.sh for deploying the registry and factory with the Stellar CLI, plus scripts/testnet-smoke.mjs for an opt-in Testnet check. The EVM repository includes deploy.sh for Base and Base Sepolia. These deploy or test infrastructure; they are not the planned creator CLI.
The documentation itself exposes llms.txt and plain-text documentation so agents can read the current product and implementation boundaries.