https://github.com/yuens1002/employer-agent

The employer side of the Open Employment Protocol — a role-agent that exposes verified claims (real stack, team, remote policy, not JD copy) and queries candidate agents to screen on signal before HR. The missing half: scaffolded as a reference and a call to build.
https://github.com/yuens1002/employer-agent

a2a agent ats employer hiring mcp oep open-employment-protocol recruiting reference-implementation role-agent verified-claims

Last synced: 17 days ago
JSON representation

The employer side of the Open Employment Protocol — a role-agent that exposes verified claims (real stack, team, remote policy, not JD copy) and queries candidate agents to screen on signal before HR. The missing half: scaffolded as a reference and a call to build.

• Host: GitHub
• URL: https://github.com/yuens1002/employer-agent
• Owner: yuens1002
• License: mit
• Created: 2026-06-02T10:53:02.000Z (about 1 month ago)
• Default Branch: main
• Last Pushed: 2026-06-02T11:54:17.000Z (about 1 month ago)
• Last Synced: 2026-06-02T13:28:21.952Z (about 1 month ago)
• Topics: a2a, agent, ats, employer, hiring, mcp, oep, open-employment-protocol, recruiting, reference-implementation, role-agent, verified-claims
• Size: 5.86 KB
• Stars: 0
• Watchers: 0
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • Contributing: CONTRIBUTING.md
  • License: LICENSE

Awesome Lists containing this project

README

          
# employer-agent — the employer side of the Open Employment Protocol

> **The missing half — scaffolded, not yet built. This repo is an invitation.**

[`resume-agent`](https://github.com/yuens1002/resume-agent) is the candidate-side reference

of the [Open Employment Protocol](https://github.com/yuens1002/open-employment-protocol): a live, queryable,

signed agent. **This is its mirror** — the employer-side reference. It does two things a

walled-garden job board never lets you do:

1. **Consume** — let a hiring *engineer* query a candidate's agent directly (`/info`,

   `/query`, `/match`, the MCP `ask_candidate` tool) and screen on real signal, in their own

   workflow, **before HR or a keyword filter is involved.**

2. **Expose** — publish a `/role` endpoint of **verified claims** (actual stack, team size,

   remote policy) — domain-co-signed, not JD marketing copy — so the candidate can verify

   *you* too. Mutual `/verify` before either side spends time.

## Status: a call, not a finished thing

Planned since 2026-04-24; **not yet built.** This scaffold exists so that someone who hires —

or builds hiring tooling — can stand it up. Per the protocol's

[evidence-graph method](https://github.com/yuens1002/open-employment-protocol/blob/main/EVIDENCE-GRAPH.md),

**building it is itself the proof the other side of the protocol is real.** A working

`/role` endpoint is worth more than any amount of agreement.

If you build it, you don't just contribute code — you become the **employer voice** the

reconciliation needs. The protocol can't be netted honestly with only one side staffed.

## Maintainership & independence

This repo is deliberately **not owned or driven by the candidate-side author.** The person

who built [`resume-agent`](https://github.com/yuens1002/resume-agent) (the candidate

reference) stays in a **mentor-only** role here and recuses from the employer-side work —

because the whole point of the protocol is that **no single party owns both sides.** Someone

who controlled the candidate side *and* the employer side would defeat the independence the

trifecta exists to protect; that conflict of interest is real, and naming it is the honest

move.

So the employer side needs a real owner — someone who hires, or builds hiring tooling, to

take it and run. **Mentorship is available; ownership is open.**

## How it plugs into OEP

OEP is a **reconciler, not a spec author**: the candidate side and the employer side each

bring their own spec, and OEP poses both and settles a net outcome no single party owns.

This repo is the **employer position made executable** — the running counterpart to

[`positions/employer.md`](https://github.com/yuens1002/open-employment-protocol/blob/main/positions/employer.md)

in the OEP repo. Build here; the standard gets reconciled there.

See **[ROLE-AGENT.md](./ROLE-AGENT.md)** for the concrete endpoint sketch to build against —

a *starting target*, not an authority (the authoritative shape is whatever the OEP

reconciliation ratifies).

## Build the smallest version

- **An afternoon** — a minimal `GET /role` returning verified claims (`actual_stack`,

  `team_size`, `remote_policy`) and a `/.well-known/agent-card.json` so a candidate agent

  can discover you. That alone completes one direction of the handshake.

- **A weekend** — add the consume side: paste a JD + a candidate agent URL, run a screening

  pass via the discovered MCP, draft grounded outreach from real evidence (not a résumé

  guess).

- **The handshake** — implement mutual `/verify`: each side confirms the other's

  domain-signed claims before revealing more. The trust mechanism is described in

  [OEP TRUST.md](https://github.com/yuens1002/open-employment-protocol/blob/main/TRUST.md).

## Provenance

If AI helps you build it, that's on-mission — value transfer, not extraction. No need to

name a tool; the working endpoint and its evidence speak.

## License

[MIT](./LICENSE) — build it, make it yours, keep the protocol no one's to own.