Interactive guide — 02 · quiz
Which agent do you need?
Each answer maps to fields the catalog already tracks: the job picks a category plus required capability tags, the surface filters on interfaces and hosting, and the leash adds autonomy requirements like long-horizon tasks or human handoff. A system scores on category fit (3), each required capability (2), each bonus capability (1), and interface/hosting matches (2).
What's the job?
Where should it live?
How much leash?
Matches
Answer the three questions — matches are scored against the live catalog.
Frequently asked
How are matches scored?
Each answer maps to fields the catalog already tracks: the job picks a category plus required capability tags, the surface filters on interfaces and hosting, and the leash adds autonomy requirements like long-horizon tasks or human handoff. A system scores on category fit (3), each required capability (2), each bonus capability (1), and interface/hosting matches (2).
What do the dashed chips mean?
Dashed 'no X' chips mark a requirement the entry doesn't have — the job needs it, the system doesn't list it. Solid chips are matched signals. 'No code execution' on an L2 result means it can take the ticket but can't run the build itself.
Why did I get frameworks instead of agents?
If you picked 'build or connect agents', the honest answer is that you need infrastructure — runtimes, protocols, tool registries — not an end-user agent. The catalog tags those separately (agent services, communication, bridges), and the quiz doesn't pretend otherwise.
Is this a recommendation?
It's a filter, not an endorsement. The quiz narrows the catalog to entries whose documented features match your requirements — compare candidates on their spec sheets, which carry pricing, status, caveats, and sources.