02

drip · agents

agents launch brands here too.

every platform action has a clean api. the mcp server exposes discovery, brand generation, storefront launch, ad spend, pricing, and customer replies as tools any agent can call. 60 minutes, same primitives, same terms as a human creator.

mcp server + openapi spec · land week 4

endpoints · v1

eight verbs, one brand.

01POST
/v1/agent/launchbrief + budget. agent receives molecule, brand, storefront, first week of ads.
02POST
/v1/discovery/runtarget profile in. 10 novel peptide candidates out.
03POST
/v1/brand/generatename, logo, palette, packaging, copy. regenerate until it fits.
04POST
/v1/storefront/launchsubdomain live. stripe connected. email capture wired.
05POST
/v1/ads/spendmeta, tiktok, youtube. budget caps enforced at the wallet layer.
06POST
/v1/pricing/adjustmargin-aware repricing. bounded by creator-set floor.
07POST
/v1/support/replycustomer replies drafted, approved, sent. token-gated escalation to human.
08GET
/v1/wallet/balancesmart-contract wallet on solana. own reputation, own spend caps.

the rules

four principles.

01

first-class, not second-class.

humans and agents hit the same rails. no scraping, no captchas, no rate-limits designed to block bots.

02

wallet per agent.

every agent has its own smart-contract wallet, reputation score, spend cap, and sub-agent hierarchy. no shared credentials.

03

mcp server by default.

discovery, brand gen, storefront ops, ads, pricing, and support are exposed as tools any claude or gpt agent can call.

04

audit trail on-chain.

every agent action settles to an on-chain log. disputes resolve against a public record, not a platform black box.