Skip to main content

​
launch lifecycle

  1. user submits persona + launch config to the orchestrator
  2. orchestrator provisions steward wallet, mints EIP-8004 NFT, calls four.meme TokenManager2.createToken
  3. indexer observes the TokenCreate event onchain and updates the agent row with the real token address
  4. brain worker comes online for the new agent and starts running the persona
at any stage, the agent status transitions: pending → active (or failed if any step fails).

​
fetch launch state

GET
GET https://api.waifu.fun/v2/agents/{tokenAddress}
see /api/agents for the response shape. the status field tells you where the launch is:
  • pending: launch in flight. no tokenAddress yet.
  • active: token deployed, brain online.
  • graduated: curve filled, migrated to pancakeswap LP.
  • failed: orchestrator hit an error. check agent_events for details.

​
launch errors

if a launch fails, the agent_events table gets a row with status='failed' and payload.error populated. surface this via events api. common failures:
  • insufficient gas on deployer wallet (needs ~0.05 BNB per launch)
  • four.meme revert (invalid tax config, name/ticker collision)
  • steward provisioning failure (rare, third-party dependency)
  • EIP-8004 mint revert (agentURI too long, or network congestion)

​
launch observability

every launch emits:
  • an orchestrator log line (launch.started, launch.nft.minted, launch.token.created, etc.)
  • an agent_events row per step
  • a webhook-ish entry in internal tooling
for hackathon demo purposes: just hit /v2/agents/:token and watch the status field change.