AI coding agents are everywhere. Claude Code, Cursor, Copilot, Windsurf — they're writing code, generating screenshots, creating documentation images, and building entire UIs. But when they need to host an image, the options are terrible.
The problem
Consumer image hosts like Imgur weren't built for programmatic use. Enterprise platforms like Cloudinary cost $89+/month and require a browser to sign up. Raw S3 is too complex. Every developer reinvents the wheel.
We wanted something different: an image host where an AI agent can register, upload, and get a CDN URL — all via API, in under a second.
How PixelVault works
Two API calls. That's it.
curl -X POST https://api.pixelvault.dev/v1/auth/register \
-d '{"email":"dev@example.com","password":"secure-pass"}'
# Upload
curl -X POST https://api.pixelvault.dev/v1/images \
-H "Authorization: Bearer pv_live_xxx" \
-F "file=@screenshot.png"
# → https://img.pixelvault.dev/proj_abc/img_xyz.png
Your image is on a global CDN immediately. 300+ edge locations. Sub-50ms latency. Zero egress fees (thanks to Cloudflare R2).
Agent-first, not agent-compatible
PixelVault isn't a traditional image host with an API bolted on. The API is the product. We ship:
- A Claude Code skill for native integration
- OpenAPI 3.1 spec for any agent to consume
llms.txtand.well-known/api-catalogfor agent discovery- Zero-friction registration (no browser, no signup form)
Free to start
The free tier includes 500 MB storage, 1,000 uploads/month, and 1 GB bandwidth. No credit card required. No trial expiry. The entire MVP runs on Cloudflare and Neon free tiers — our infrastructure cost is literally $0.
When you need more, Starter starts at $9/month.
What's next
We're launching with upload, serve, list, and delete. Coming soon: custom domains, auto-optimization (WebP/AVIF), private images with signed URLs, and skills for more coding agents.