Expand description
RLM Router — decides when and how to route content through RLM.
Splits routing decisions, smart truncation, and the iterative auto-processing loop across focused submodules.
Structs§
- Crate
Auto Process Context - Internal context for the crate’s auto-process loop.
- Process
Progress - Progress tick during auto-processing.
- Routing
Context - Context for a routing decision.
- Routing
Result - Outcome of a routing decision.
Enums§
- Host
Tool Result - Result of dispatching a single RLM tool call.
Traits§
- Into
Crate Ctx - Trait for converting a host context into the crate’s internal form.
- Router
Host - Narrow interface the router needs from the host environment.
Functions§
- auto_
process - Process large output through the RLM iterative analysis loop.
- extract_
final - Extract the answer from a
FINAL("…")call in LLM text. - fallback_
result - Build a fallback
RlmResult(truncation only, no LLM call). - should_
route - Decide whether tool output should be routed through RLM.
- smart_
truncate - Truncate output to
max_tokenswith head/tail preservation.