The common language for AI agents, talent platforms, employers, and recruiters.
UHP is an open-source protocol that standardizes how AI agents and systems interoperate to enable discovery, application, screening, interviewing, and hiring workflows — with privacy first, intent-based actions, and secure consent mechanisms built into the core.
Hiring today is fragmented across job boards, ATS systems, HR APIs, and recruiter tools. This creates friction for AI assistants, multiplatform agents, and integrated workflows.
UHP provides a shared protocol that:
- Enables AI agents (e.g., chatbot assistants, LLM-based job search copilots) to discover roles, apply, schedule interviews, and negotiate offers using a common interface
- Embeds privacy controls and progressive disclosure by default
- Reduces custom point-to-point integrations between candidate systems and employer systems
- Supports intent-based actions, not just CRUD APIs
- Allows systems to interoperate without exposing unnecessary personal data
Explore the Protocol Overview, Privacy Model, and Specification to understand:
- Core concepts: Actors, intents, states
- Canonical JSON schemas for jobs, candidates, applications, and consent
- How visibility levels and purpose-bound access govern data exposure
- Agent-native workflows for application and interview lifecycle
UHP is inspired by open standards like the Universal Commerce Protocol (UCP) — designed for agentic interoperability in commerce — but built for hiring and recruitment with privacy as a first-class citizen.
Key principles include:
- Standardized Interoperability: Uniform way for platforms, agents, and systems to interact, regardless of backend systems
- Privacy-Native: Strong visibility levels, consent lifecycles, and purpose-bound data access
- Agentic Hiring: Intent-driven actions and state machines make it easy for autonomous AI agents to drive workflows
- Extensible: Schema and capabilities can be extended for industry-specific or advanced use cases
This repository contains:
- The UHP protocol specification
- JSON Schemas for core objects (Job, Candidate, Application, Consent)
- Reference definitions for intent actions and state machines
- Examples and best practices for integrating UHP with AI agents
See the /specification directory for full details.
UHP treats privacy as a core building block, not an afterthought:
- Visibility levels (Public, Anonymized, Restricted, Sensitive) govern what data can be exposed at each stage of the hiring workflow
- Consent objects encode state and purpose bindings
- Data access is purpose-bound — agents must declare why they need access to specific fields
- Designed to be compatible with real-world privacy regulations
Unlike traditional HR APIs, UHP is designed to be interpretable and reasoned over by AI agents:
- Intent-based action envelope (
ApplyForJob,ScheduleInterview, etc.) - Explicit state machines for tracking workflow progression
- Machine-readable capability discovery (see Capability Discovery Endpoint)
- Privacy-aware consensual access patterns
- Clone the repository
- Review the JSON Schema definitions in
/schemas - Read the protocol overview in
/docs - Use intent action examples in
/examplesto simulate agent workflows - Integrate with your platform or agent implementation
We actively welcome contributions — from schema improvements to implementation examples:
- Found a bug? Raise an issue.
- Want to propose a new capability? Write a proposal in
/proposals. - Interested in governance or extensions? Join the discussion.
Together, we can build the standard protocol for the future of hiring interoperability.
UHP is released under the Apache License 2.0.
For inspiration and background on agent-centric protocols in adjacent domains, see:
- Universal Commerce Protocol (UCP) — an open standard for agentic commerce interoperability, enabling discovery and transactions across platforms and AI assistants. :contentReference[oaicite:0]{index=0}