I can’t set the effort when configuring a sub-agent! How can I configure this?

[j-token]

Question

Recently, I’ve been using OpenAI, and I use it with a custom model added to OpenAI.
openai/gpt-5.2-codex-high

The reason I use it this way is because sub-agents cannot be configured with Ctrl + T.
This issue started occurring after the recent update, but there is no documentation explaining how to handle or resolve it. What should I do?

[github-actions]

This issue might be a duplicate of existing issues. Please check:

• [FEATURE] Support default variant configuration per agent (#7156) #7138: Support default variant configuration per agent - Proposes adding a variant field to agent configuration to set default effort levels without needing Ctrl+T
• [FEATURE]: Support for Model Variants in (sub)agents #6645: Support for Model Variants in (sub)agents - Requests variant specification for sub-agents to avoid errors when switching between thinking and non-thinking models
• [FEATURE]: Commands Model Variant Specification #7713: Commands Model Variant Specification - Extends variant configuration to commands

All three address the core problem: sub-agents cannot be configured with Ctrl+T, and variants cannot be set in the configuration file. The proposed solution in #7138 would allow you to configure effort levels directly in your agent config.

Feel free to ignore if none of these address your specific case.

[j-token]

i found to way out

in agent/test.md

---
model: openai/gpt-5.2-codex
mode: subagent
options:
  reasoningEffort: high
  reasoningSummary: auto
  textVerbosity: medium
---

it works....!

[mugnimaestra]

This should get easier soon — PR #12567 makes subagents automatically inherit the parent session's thinking variant when they don't have their own configured. So if you set high on your main session, Task-spawned subagents will use high too, without needing per-agent configuration.

[github-actions]

To stay organized issues are automatically closed after 90 days of no activity. If the issue is still relevant please open a new one.