Skip to content

What Is Open Job Context Protocol (OJCP)?

The Open Job Context Protocol (OJCP) is an open standard for how AI agents search for jobs, read job details and apply on a candidate's behalf, under rules the employer or job board sets. Built on the Model Context Protocol (MCP), it requires the candidate's consent before an agent submits an application and lets a job board demand proof of which agent is applying.

AI & Machine Learning in Recruitmentojcpai-agentsmcpjob-boardsopen-standardUpdated November 2026

Why OJCP Matters for Job Boards

Candidates now send AI agents to look for work and apply for them. Recruitics' launch announcement describes agents clicking through job descriptions, typing information into career sites and filling HR inboxes with applications, and it says most of those submissions are low quality or mismatched because the agent has no reliable way to judge fit before applying.

OJCP replaces that guesswork with a shared format. A job board publishes its jobs, its apply options and its rules once, and any agent that speaks OJCP can work with it, without a one-off integration on either side.

How OJCP Works

OJCP tools are standard MCP tools, so any MCP client can call them. OJCP calls any service that exposes jobs this way a provider. A provider describes itself in a manifest at /.well-known/ojcp.json, listing its tools, apply paths and authentication rules, and agents find providers by checking that file or by querying the OJCP registry, a public list with one reviewed entry per provider.

The spec defines six standard tools. search_jobs is the only one a provider must implement, while get_job_detail, get_employer_context, begin_application, submit_application and check_application_status are recommended. Job data extends the schema.org JobPosting type, the same vocabulary Google uses for job posting structured data, and search results can include a fit_score and fit_rationale when the agent passes a consent-scoped candidate profile.

An application follows four steps. The agent discovers the provider, searches and evaluates jobs, starts the application with begin_application by passing an agent declaration and the candidate's consent token, and then tracks progress with check_application_status.

OJCP is designed so a job board can tell an accountable agent from an anonymous scraper. Agents declare who they are and who they act for when they start an application, and they sign their requests with HTTP Message Signatures on the Web Bot Auth profile, so the provider can check which software is calling without a central gatekeeper.

A verified identity proves which software is calling, not that a person agreed to apply. That is why OJCP enforces a consent gate before submission, and why it defines the user mandate: a credential issued under the user's authority that binds the agent's signing key to one specific job, application and set of candidate data, and that can be used once.

For roles that need verified human identity, such as finance, government and healthcare, OJCP works with third-party verifiers like ID.me and CLEAR. The proof can be collected inside the provider's apply form or by the agent, and in both cases the provider validates a cryptographic proof while no raw personal data passes through the protocol.

What OJCP Means for a Staffing Agency's Job Board

For a staffing agency, OJCP is a way to make open roles readable by AI agents without rebuilding the job board. Recruitics' announcement says OJCP layers on top of existing infrastructure and that job boards can start by publishing a discovery file. Because the spec requires only search_jobs, a read-only provider can answer agent searches before any apply flow exists.

The agency keeps control of what agents see and do. A board can mark individual jobs as public, restricted to specific verified agents, or private to partner feeds, and it can require a verified agent identity for the requests it chooses. Gating depends on the agent, never on the candidate. Each application is credited to the source that surfaced the job, using evidence the board already holds.

Before going live, an implementation can be checked against the public conformance suite, and a provider is added to the registry by pull request.

Who Builds and Governs OJCP

Recruitics announced OJCP on July 29, 2026, and Recruitics CTO Austin Anderson authored the spec. A nine-seat founding steering committee now governs the protocol. Its members share equal authority and none holds a veto.

OJCP is free to use. The specification, schemas and docs are licensed under CC BY 4.0 and the code and tooling under Apache 2.0, with everything published at github.com/ojcp-org. The current release, 0.3, came out on October 8, 2026, and minor releases are additive and backward compatible.

Related Terms

Frequently Asked Questions

Is OJCP free to use?
Yes. Recruitics describes OJCP as a shared standard that is free to use and collectively operated. The specification is licensed under CC BY 4.0 and the code under Apache 2.0, and the full spec, a reference implementation and a live testing environment are available at ojcp.dev.
What is the minimum a job board needs to support OJCP?
A manifest at /.well-known/ojcp.json and the search_jobs tool. search_jobs is the only tool the spec makes mandatory, and the other five standard tools, including begin_application and submit_application, are recommended. The ojcp-org conformance suite checks an implementation against the published schemas.
How is OJCP related to MCP?
OJCP is built on the Model Context Protocol, which Recruitics' announcement calls the emerging standard for how AI agents talk to tools. OJCP tools are valid MCP tools, so any MCP client can call search_jobs, get_job_detail and begin_application. OJCP adds the jobs layer on top: job and employer schemas, apply paths, agent identity, consent and verification.
Who uses OJCP today?
As of November 2026, the project's public adopters list shows three live providers: the Recruitics reference provider at ojcp.dev, FoundRole and freehire, with reqspace.ai listed as evaluating. Organizations add themselves to that list in good faith, and the project does not formally verify the entries.

Sources

  1. [1]OJCP official site
  2. [2]OJCP specification repository (GitHub)
  3. [3]OJCP adopters list (ADOPTERS.md)
  4. [4]OJCP provider registry (GitHub)
  5. [5]OJCP conformance suite (GitHub)
  6. [6]Recruitics launch announcement, July 29, 2026 (EIN Presswire)
  7. [7]Google Search Central: Job posting structured data