Ask research questions and get answers grounded in sources that automatically pay their creators.
Register your own articles so an AI agent can cite and pay you when it uses them.
Reuse the research agent, wallet API wrapper, and treasury code to build your own pay-per-call AI service.
Track live payment activity and citation earnings through the dashboard.
| vickman787/citeflowai | ajr-khll/ghostty-config-cli | audemodo/audemodo-responsive-keepalive | |
|---|---|---|---|
| Stars | 28 | 28 | 28 |
| Language | TypeScript | TypeScript | TypeScript |
| Setup difficulty | moderate | easy | easy |
| Complexity | 4/5 | 2/5 | 2/5 |
| Audience | developer | developer | developer |
Figures from each repo's GitHub metadata at analysis time.
Requires API keys for Supabase, Gemini, Anthropic, and Circle Web3 Services before it runs.
CiteFlow AI is a research assistant that answers questions and pays content creators automatically whenever it uses their work to build an answer. It was built to solve a specific problem: when an AI agent reads and summarizes someone's writing, the original writer usually gets nothing in return. This project tries to fix that by attaching real, tiny payments to each answer. Here is how the payment side works. A user connects a Circle wallet and locks a small budget upfront, for example fifty cents, before asking a question. As the AI researches an answer, it checks which registered articles it actually drew from. Each time it uses a specific creator's work, it sends that creator a very small payment, for example five cents, straight from the locked budget. The service itself takes a small fee to cover the cost of running the AI. If a question turns out to be simple and no paid sources were needed, the AI refunds the entire budget back to the user automatically. On the technical side, logging in feels like an ordinary email signup, but behind the scenes it creates a Circle wallet and links it to a Supabase account, so users get a familiar login experience while still holding their own wallet. Research history and answers are saved per user and follow them across devices. For generating answers, the system primarily uses Gemini 2.5 Flash, and automatically switches to Claude Haiku through the Anthropic API if it hits a rate limit, so the service keeps working even under load. It also uses retrieval and vector search through Supabase to find relevant registered sources quickly. The project exposes reusable pieces of code, including a research agent script, a wrapper for Circle's wallet API, and a treasury module that manages the budget and refund logic, so other builders can adapt the same pay-per-question pattern for their own AI tools. It is written in TypeScript, built on Next.js, and released under the MIT License.
An AI research assistant that pays content creators small automatic royalties whenever it uses their work in an answer.
Mainly TypeScript. The stack also includes TypeScript, Next.js, Supabase.
Use, modify, and distribute freely, including for commercial purposes, as long as you keep the copyright notice.
Setup difficulty is rated moderate, with roughly 30min to a first successful run.
Mainly developer.
This repo across BitVibe Labs
Don't trust strangers blindly. Verify against the repo.