Changelog
Improved

Browse job types without an API key

GET /jobs/types and its per-type schema endpoint are now public, so an agent or a client library can read the live job registry before anyone has signed up for a key.

API MCP

Discovery used to need a credential, which meant a tool had to know what Rendobar could do before it could ask. Both discovery endpoints are open now:

Terminal window
curl https://api.rendobar.com/jobs/types
curl https://api.rendobar.com/jobs/types/ffmpeg/schema

The first returns every active job type with its tag, summary, and the media kinds it accepts. The second returns connector-neutral field descriptors for a single type, the same ones the no-code connectors render their forms from.

Why it matters for agents

This matters most for MCP clients that start without a key. @rendobar/mcp carried a small hardcoded list of job types for exactly that case, and anything that launched the server just to read its tool list, directory indexers included, saw that list rather than the real registry. Now the keyless path reads the same registry every other caller does, so a new job type shows up everywhere the moment it ships.

Compatibility

Nothing else changed. Both endpoints still return active types only, so in-progress work stays hidden, and both still send Cache-Control: max-age=300 so repeat calls hit the edge rather than the origin. Anonymous traffic is limited to 60 requests a minute per IP. Authenticated calls behave exactly as before and keep their per-plan limits.

import { createClient } from "@rendobar/sdk";


// No apiKey. /jobs/types is public.
const types = await createClient({}).jobs.types();

The registry is the same one the API validates against, so a client that reads it sees exactly the job types the platform will accept, including ones added after the client shipped.

The same registry backs the MCP server and the SDK.

Share