Skip to main content

​
Slugs, not ids

Jobs and companies are addressed by slug — a lowercase, hyphenated string — never a numeric id.
Get a job slug from search_jobs. Get a company slug from search results, or from the company directory. Passing an unknown slug returns an error that names the problem and points at where to find valid values, rather than an empty result you might mistake for “nothing is open”. Every job a tool returns carries two URLs, deliberately:

apply_url

The untouched posting on the company’s own ATS. This is where people apply. WorkWay never sits between the applicant and the employer.

workway_url

The role’s page on workway.dev — useful for company context, related roles and salary data.

​
Resources

Alongside tools, the server exposes three MCP resources carrying reference material an assistant can read directly:
Not every MCP client surfaces resources. The get_workway_info tool returns the same material, so an assistant always has a way to reach it.

​
Free and Pro

A free account can use every tool: searching, saving jobs, following companies and editing a talent profile all work with no paid plan. Pro adds one thing — instant email alerts. When a company you follow posts a new role, Pro accounts get an email straight away rather than finding out later.
follow_company is never plan-gated. The follow itself always succeeds on every plan; only the email delivery requires Pro. The tool’s response says so explicitly on the free plan, so nobody is left assuming alerts are switched on when they are not.
list_alerts reports email_alerts_active for the same reason — it is always unambiguous whether emails will actually arrive.

​
Errors

Tool errors are written to be actionable rather than terse:
Filter validation is delegated to the same service that powers GET /api/job/list, so the MCP surface and the website can never disagree about what a valid domain, level or source is.