fix: repair live model inference edge cases

Fix live model inference edge cases across provider streaming, model switching, outbound delivery, and gateway tool resolution.

Includes live/provider issue fixes and leaves #89100 explicitly partial for the remaining FM-2 group routing case.
This commit is contained in:
Peter Steinberger
2026-06-01 23:03:27 -04:00
committed by GitHub
parent 3128ec9858
commit 9ead0ae921
76 changed files with 2704 additions and 216 deletions

View File

@@ -141,7 +141,7 @@ export function registerOnboardCommand(program: Command): void {
.option("--custom-provider-id <id>", "Custom provider ID (optional; auto-derived by default)")
.option(
"--custom-compatibility <mode>",
"Custom provider API compatibility: openai|anthropic (default: openai)",
"Custom provider API compatibility: openai|openai-responses|anthropic (default: openai)",
)
.option("--custom-image-input", "Mark the custom provider model as image-capable")
.option("--custom-text-input", "Mark the custom provider model as text-only")
@@ -217,7 +217,11 @@ export function registerOnboardCommand(program: Command): void {
customApiKey: opts.customApiKey as string | undefined,
customModelId: opts.customModelId as string | undefined,
customProviderId: opts.customProviderId as string | undefined,
customCompatibility: opts.customCompatibility as "openai" | "anthropic" | undefined,
customCompatibility: opts.customCompatibility as
| "openai"
| "openai-responses"
| "anthropic"
| undefined,
customImageInput:
opts.customTextInput === true
? false