feat: route memory LLM completions through the host model runtime (opt-in transport)#947

Draft

gorkem2020 wants to merge 12 commits into

CortexReach:masterfrom

gorkem2020:feat/runtime-llm-completions

Commits on Jul 18, 2026

feat(llm-client): add a host-managed completion transport alongside the direct client
gorkem2020
committed
build: recompile dist for the runtime-llm-completions transport layer
gorkem2020
committed
fix(llm-client): normalize the model string on the host->direct fallback path
gorkem2020
committed
fix(llm-client): forward an explicit reasoning effort on the host transport

gorkem2020
and
claude
committed
build: recompile dist for the reasoning-effort forwarding fix

gorkem2020
and
claude
committed
fix(llm-client): send reasoning effort on direct requests, dedupe fallback warn, and stop inheriting embedding credentials on host fallback
gorkem2020
committed
feat(llm-client): rename llm.reasoningEffort to llm.thinkLevel, keep deprecated alias
gorkem2020
committed
fix(llm-client): stop the llm.thinkLevel/reasoningEffort schema defaults from silently overriding explicit config
gorkem2020
committed
build: recompile dist for the thinkLevel schema-default fix
gorkem2020
committed
fix(llm-client): remove the llm.reasoningEffort legacy alias entirely
gorkem2020
committed
chore: drop the deleted alias test from the CI chain and rebuild dist

gorkem2020
and
claude
committed
chore(dist): rebuild on current master
gorkem2020
committed