Cloudflare Open-Sources pvcli Debugger for Privacy Protocols
The curl-like command-line tool lowers the barrier for debugging split-trust infrastructure used by Apple and Microsoft.
Cloudflare has released pvcli (privacy-client), an open-source command-line tool for debugging privacy-preserving protocols, making troubleshooting accessible beyond tech giants who previously could afford custom tooling.
Split-Trust Debugging
Privacy protocols like Oblivious HTTP (OHTTP) split trust across multiple parties so no single operator can link a user's identity to their activity. The relay knows the user but not the destination; the gateway knows the destination but not the user. This architecture protects privacy but makes debugging nearly impossible because no single operator has full visibility into the request flow.
"Debugging privacy protocols where different operators need to operate different pieces of the protocol infrastructure is extremely difficult," said Fisher Darling, systems engineer at Cloudflare. "It's a lot of manual work, custom scripts, and trying to troubleshoot without much information."
OHTTP powers services including Apple's iCloud Private Relay and Microsoft's Edge Secure Network, both built in partnership with Cloudflare. Before pvcli, only companies with those resources could build the necessary troubleshooting infrastructure.
Curl-Like Interface for Humans and Agents
The tool's command syntax is modeled after curl to make it intuitive for both humans and AI agents. "Pvcli was explicitly designed for agent-based debugging," Darling said. "It just so happens that humans also appreciate an interface that feels familiar to curl's."
Developers can test against their own infrastructure or a public sandbox at ohttp.info. The tool currently debugs OHTTP; support for MASQUE is planned for the future.
Lowering Barriers for Regulated Industries
By open-sourcing pvcli under the Apache-2.0 License, Cloudflare lowers the barrier for smaller companies—especially those in highly regulated sectors like health and finance—to implement advanced privacy-preserving infrastructure. The tool was tested with an unnamed set of customers during a beta phase prior to release.
The move also addresses an emerging challenge: running AI inference on personal data without leaking that information to the provider. By designing the tool for AI agents, Cloudflare enables autonomous systems to construct and debug privacy-preserving requests without human intervention.
The tool is available on GitHub.