You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
AgentDeck deliberately owns no execution. An Agent compiles to an OpenAI Agents SDK agent, a Workflow compiles to a LangGraph graph, and the engine adapter is the seam between the two.
Adding an engine means writing an adapter, not rewriting the runtime.
So: which engine would actually change what you can ship?
Candidates that have come up:
PydanticAI — closest in spirit to the typed-declaration side of AgentDeck
Agno — overlaps with the runtime story, which makes the boundary interesting
Bare Chat Completions / Responses — no framework at all, just the runtime around it
CrewAI / AutoGen — multi-agent orchestration with their own opinions
something else
What is most useful in a reply is not the name but the constraint behind it: what you are running
today, why you can't move it, and which part of AgentDeck you want around it. An adapter is only
worth writing where the runtime half genuinely earns its place.
No adapter is promised here — support ships when it is real, not when it is announced.
reacted with thumbs up emoji reacted with thumbs down emoji reacted with laugh emoji reacted with hooray emoji reacted with confused emoji reacted with heart emoji reacted with rocket emoji reacted with eyes emoji
Uh oh!
There was an error while loading. Please reload this page.
AgentDeck deliberately owns no execution. An
Agentcompiles to an OpenAI Agents SDK agent, aWorkflowcompiles to a LangGraph graph, and the engine adapter is the seam between the two.Adding an engine means writing an adapter, not rewriting the runtime.
So: which engine would actually change what you can ship?
Candidates that have come up:
What is most useful in a reply is not the name but the constraint behind it: what you are running
today, why you can't move it, and which part of AgentDeck you want around it. An adapter is only
worth writing where the runtime half genuinely earns its place.
No adapter is promised here — support ships when it is real, not when it is announced.
All reactions