Thanks everyone for helping land #27380.

Follow-up goal:

• Keep the fix(onboard): increase verification timeout and reduce max_tokens for custom provider probes #27380 verification behavior, and add endpoint-aware timeout policy using shared network helpers (not bespoke URL classification in onboard code).

Why:

• Local/self-hosted endpoints can have slower first-response behavior than remote hosted APIs.
• We want robust onboarding verification without introducing one-off host parsing logic.

Scope:

• Refactor verification timeout selection in src/commands/onboard-custom.ts to use shared network helper utilities.
• Preserve current lean verification behavior from fix(onboard): increase verification timeout and reduce max_tokens for custom provider probes #27380.
• Add/expand tests for host categories:
  • localhost
  • 127.0.0.1
  • RFC1918 addresses (for example 172.17.x.x)
  • host.docker.internal

Acceptance criteria:

• Timeout policy decision is implemented via shared helper(s), not ad-hoc URL classification in onboarding command logic.
• Existing fix(onboard): increase verification timeout and reduce max_tokens for custom provider probes #27380 verification behavior remains intact.
• Test coverage explicitly includes the host categories above.

Related:

• fix(onboard): increase verification timeout and reduce max_tokens for custom provider probes #27380
• fix: increase VERIFY_TIMEOUT_MS to 2 minutes for local LLMs #29062
• [Bug] Onboard wizard fails to verify local llama.cpp server (Custom Provider) — but API works fine from container #27346
• [Bug]: VERIFY_TIMEOUT_MS in onboard-custom.ts is too short for local LLMs. #28972