We want you to build the next Git platform on Cloudflare
Source
blog.cloudflare.com
Date
Context
Cloudflare's Artifacts gives developers programmable Git repositories for AI agentAn AI system that doesn't just answer once but works toward a goal in a loop — taking actions, reading the results, and deciding what to do next.Full definition → work. In its open-beta announcement, a Worker can create or fork a repository, inspect its files and issue access tokenThe chunk of text a model reads and writes in — roughly three-quarters of a word — and the unit AI usage is billed in.Full definition → scoped to that repository.
This supports a concrete workflow: give an agent a separate copy for a task, then react when it pushes changes. Repository events can trigger a review or continuous-integration workflow. Artifacts supplies the storage and event mechanisms; developers still define how competing changes are reviewed and merged. The accompanying competition asks builders to develop that coordination layer.
Key takeaways · AI-distilled
Artifacts repositories can now deploy to Workers through Workers Builds: pushes to the production branch build and deploy the Worker, while pushes to other branches create or update isolated, shareable Workers Previews.
An Artifacts binding makes Git programmable from a Worker: code can create or fork repos, read files and commits, and issue repo-scoped Git tokens. Cloudflare's example forks a project for a new agent task and reads its AGENTS.md for instructions.
Artifacts emits events when a repo is created, imported, forked, deleted, pushed to, cloned or fetched. Subscribing to push events lets a Worker start a review workflow for each push, passing the repo, branch and new commit to a review agent.
Namespaces can be pinned to U.S. or EU jurisdiction for storage and processing, and per-repo metrics show operations, pulls, pushes, errors and error rate. Billing for repository operations and stored data starts October 15, 2026.
Cloudflare's competition wants an agent-native Git platform on Workers and Artifacts, not GitHub with agents added. Entries need concurrent multi-agentUsing several AI agents on one problem — splitting work in parallel, checking each other, or filling different roles like planner and reviewer.Full definition → changes, permissively licensed source and a 5-10 minute demo by October 14, 2026; first place gets $25,000 in credits.
Terms in this piece · Glossary
AI agent — An AI system that doesn't just answer once but works toward a goal in a loop — taking actions, reading the results, and deciding what to do next.
token — The chunk of text a model reads and writes in — roughly three-quarters of a word — and the unit AI usage is billed in.
multi-agent — Using several AI agents on one problem — splitting work in parallel, checking each other, or filling different roles like planner and reviewer.
