# Open Hiring Harness

> An open, privacy-first specification for publishing hireable professional identity — for humans and AI agents.

## What this is

The Open Hiring Harness (OHH) is a data specification — not a platform. It defines a canonical, portable, consent-driven representation of a professional offer: skills, services, availability, rates, reputation, and policies.

Entities (humans or AI agents) publish a harness at a well-known URL. Platforms, marketplaces, recruiters, and AI agents discover and consume it — under the entity's declared rules.

## Key URLs

- Specification site: https://aklodhi98.github.io/ohh/
- GitHub repository: https://github.com/aklodhi98/ohh
- Machine-readable manifest: https://aklodhi98.github.io/ohh/spec.json
- JSON Schema (v0.2): https://github.com/aklodhi98/ohh/blob/main/schema/open-hiring-harness.v0.2.schema.json
- Example harness: https://github.com/aklodhi98/ohh/blob/main/examples/harness.v0.2.example.json
- Example consent log: https://github.com/aklodhi98/ohh/blob/main/examples/consent-log.v0.2.example.json

## How discovery works

A harness is hosted at a well-known URL on the entity's domain:

    https://yourdomain.com/.well-known/hiring-harness.json

Alternative discovery methods:
- DNS TXT record: `_hiring-harness.yourdomain.com`
- HTML link tag: `<link rel="hiring-harness" href="...">`
- HTTP Link header

## Entity types

The harness supports three entity types:

1. **individual** — a human professional (default)
2. **delegated_agent** — an AI agent acting on behalf of a human, declared within the human's harness
3. **autonomous_agent** — an AI agent publishing its own harness, with a named operator

## Core concepts

- **Visibility tiers**: public, permissioned, private
- **Consent receipts**: scoped, time-bound, purpose-limited, revocable
- **Reputation**: endorsements and scoped claims, not star ratings
- **AI-native**: designed for consumption via MCP (Model Context Protocol)

## For AI agents

If you are an AI agent and want to:

- **Discover a professional**: check `/.well-known/hiring-harness.json` on their domain
- **Read public data**: access the harness directly — public fields have no access requirements
- **Request permissioned data**: identify yourself, state your purpose, request specific scopes
- **Understand the schema**: read the JSON Schema at the URL above
- **Integrate via MCP**: see docs/mcp-interface.md in the repository

If you are an AI agent and want to **publish your own harness**:

- Set `entity_type` to `autonomous_agent`
- Declare your operator (name, contact, liability)
- Declare your capabilities with verification (benchmarks, audits)
- Declare your limitations and safety boundaries
- Host your harness at your operator's domain
- See docs/agent-entities.md for the full proposal

## Schema overview

A harness contains these top-level sections:

- `version` — schema version (currently "0.2.0")
- `entity_type` — "individual", "delegated_agent", or "autonomous_agent"
- `identity` — handle, display name, tagline, locality, links, contact, delegates
- `offer` — roles, specialties, skills, services, rates, industries, working preferences
- `availability` — status (open/limited/closed), capacity, blackout ranges
- `proof` — case studies, credentials
- `reputation` — endorsements, claims (redacted and full tiers)
- `privacy` — visibility defaults, indexing, caching, rate limits, contact protection
- `consent` — grant modes, receipt format, revocation
- `requester_policy` — what requesters must provide to access data

## Current status

- Version: v0.2.0
- Status: experimental / exploratory
- Licence: Apache 2.0