offload_middleware

Stable prefix for forced-compaction failure tool messages.

/offload drives the tool server-side and can only observe the resulting ToolMessage text across the LangGraph server boundary, so it keys failure detection on this prefix. Owning the literal here means the producer (_forced_compact_error) and both consumers (app._drive_server_side_compaction live-stream detection and app._find_compaction_failure committed-state scan) reference one constant instead of re-hardcoding the wording independently.

Note: this value is deliberately identical to the leading text of the SDK's own model-initiated compaction-failure message, so a failure emitted by either path is recognized. Because the scan is bounded to messages produced by the current /offload attempt, a stale failure from an unrelated prior turn is not matched. Only the prefix position is load-bearing; wording after it is free to change.

Declared context_schema for the agent graph.

Registered via context_schema= when the graph is built, so LangGraph coerces each run's context= payload into this dataclass — in-process, runtime.context is a CLIContextSchema instance.

It exists alongside CLIContext (below) because the payload is shaped differently on each side of the API boundary: in-process it is coerced to this dataclass, but over the LangGraph API server (RemoteGraph) it is serialized to JSON and arrives as a plain dict. Consumers (configurable_model._get_context, _should_interrupt_tool_call) therefore accept both shapes. CLIContext is the client-facing builder for constructing that payload.

Fields mirror CLIContext; see its per-field docstrings for semantics.

Active model configuration read from a tool runtime.

A named tuple rather than a bare 4-tuple so the two structurally identical dict slots (model_params, profile_overrides) are addressed by name at both the construction sites (keyword args) and the read site (attribute access) — a silent positional transposition the type checker would not catch is thereby avoided. Positional construction/unpacking is still possible and would defeat this, so call sites must keep using names.

Add explicit forced compaction and runtime model selection for dcode.

The SDK tool's normal, model-initiated behavior remains unchanged. The private force input is used only by the user-initiated /offload path, which must compact whenever messages exceed the retention window even when the conversation has not reached the SDK's proactive eligibility gate.