All updates

Public API, CLI, and MCP server

Cavuno now ships a public REST API, a `cavuno` command-line tool, and a remote MCP server for AI clients. Read and write jobs and companies from scripts, terminals, or Claude, Cursor, and Windsurf.

Abi Tyas TunggalAJack WalshJ
By Abi Tyas Tunggal and Jack Walsh on

Programmable access to your job board, on every interface developers and AI agents already use. The same surface area is available three ways: REST, CLI, and MCP.

REST API

  • Versioned at api.cavuno.com/v1: stable endpoints for jobs, companies, settings, and media, with full OpenAPI spec
  • Stripe-style responses: flat envelopes ({"object": "list", "data": [...], "has_more": true}), cursor-based pagination, and Stripe-voice docs across every field
  • OAuth 2.1 with dynamic client registration: standards-compliant authorization server, JWKS endpoint, per-request consent, and RFC 8252 loopback support for native apps
  • Idempotency keys: safely retry mutating requests; replays return the original response with an Idempotency-Replayed marker
  • Audit log endpoint: every API key and OAuth action shows up at /v1/audit-logs with actor, IP, and resource trail
  • Reference docs: browsable API reference with code samples in every language

Cavuno CLI

  • npm install -g cavuno: scripted access from any terminal or CI pipeline
  • Full surface: jobs list, jobs create, jobs publish, companies create, settings update, and more — every API endpoint has a matching command
  • Auto-generated docs: /docs/cli is generated from the CLI's command tree, so the website never drifts from the binary
  • CSV-friendly: pipe output into jq or csvkit for batch operations

MCP server for AI clients

  • Remote MCP server: connect Claude, Cursor, Windsurf, or any MCP-compatible client and let them read and write jobs and companies on your behalf
  • Same surface as the API: every API endpoint is exposed as a CLI command and an MCP tool, so an AI agent can do whatever a developer can do