Live
Tool access
Discovery and binding run end to end through the governed gateway, with every supported framework; see Framework binding.
Governed tool access lets an agent discover the MCP tools your organisation has published and governed, filter them down to what it actually needs, and bind them into any of the eight supported frameworks as that framework’s native tool objects — the same “no wrapper” approach as model access.
Two lines from catalog to agent
tools = await donkey.tools.discover(tags=["approved"])
agent = create_react_agent(donkey.langgraph.chat_model("gpt-4o"), await tools.langgraph())Everything else in this section — search filters, session management, per-framework binding and lockfile pinning — makes those two lines hold up in production.
Outbound A2A is separate: donkey.a2a.send() and
context-managed stream() use official SDK objects. They are offline-verified;
successful gateway send/SSE is not yet live-verified by this implementation,
and framework-native agent-as-tool binding remains planned.
Discover and filter
donkey.tools.discover(...) is also the filter entry point. One call narrows
the catalog by governance and tags, so an agent binds only the tools it
needs:
tools = await donkey.tools.discover(
tags=["approved"],
governed=True, # the default: default criteria; or an explicit GovernanceCriteria
)Name search and asset type filters live one layer down, on
donkey.registry.search(), which is live.
“Governed” is a computed, environment-scoped predicate rather than a flag in Exchange — see Discovery, search & filter.
In this section
Narrow the catalog by governance and tags, or search it by name and asset type.
Framework bindingTurn a ToolSet into each framework’s own native tool objects.
Pin each server’s listed tools in donkey.lock and fail on drift.
Send SDK requests and stream SDK events through a registry-pinned gateway. Native framework binding remains planned.
Offline-verified