[Question]: Is the Hephaestus model's implementation hard-coded, or was it designed this way intentionally?

[Eric-GoodBoy-Tech]

Prerequisites

• I will write this issue in English (see our Language Policy)
• I have searched existing issues and discussions
• I have read the documentation or asked an AI coding agent with this project's GitHub URL loaded and couldn't find the answer
• This is a question (not a bug report or feature request)

Question

Is the Hephaestus model's implementation hard-coded, or was it designed this way intentionally?My configure did't work.

Context

No response

Doctor Output (Optional)

Question Category

Performance

Additional Information

No response

[acamq]

Thank you for raising this issue. I've investigated the Hephaestus agent configuration system, and the model is NOT hard-coded. Hephaestus uses a flexible model resolution pipeline that respects user configuration.

How Hephaestus Model Resolution Works

The model resolution follows this priority order:

1. User override (via config agents.hephaestus.model)
2. Category default (if agents.hephaestus.category is set)
3. Provider fallback chain (default: gpt-5.3-codex → gpt-5.4)
4. System default (ultimate fallback)

Important Constraint

There is a built-in hook (no-hephaestus-non-gpt) that blocks Hephaestus from using non-GPT models by default. This is intentional design—Hephaestus is optimized specifically for GPT models (GPT-5.3 Codex / GPT-5.4).

If you try to use Hephaestus with Claude, Kimi, GLM, etc., the system will:

• Show an error toast
• Automatically redirect to Sisyphus instead

To Help Diagnose Your Issue

Could you please provide:

1. Your configuration file (.opencode/oh-my-opencode.jsonc or ~/.config/opencode/oh-my-opencode.jsonc) — specifically the agents.hephaestus section

2. What model are you trying to use with Hephaestus? (e.g., Claude, Kimi, GLM, a specific GPT variant)

3. What providers do you have connected? Run opencode doctor or check your provider setup

4. What behavior are you seeing? — Is Hephaestus not appearing at all, or is it using a different model than configured?

If You Want to Use a Non-GPT Model with Hephaestus

Add this to your config:

{
  "agents": {
    "hephaestus": {
      "model": "your-preferred-model",
      "allow_non_gpt_model": true
    }
  }
}

However, I'd recommend using Sisyphus for non-GPT models, as Hephaestus's prompt is specifically tuned for GPT's coding strengths.

---

Summary: Not hard-coded, but there are intentional design constraints. Once you provide the details above, I can help identify exactly why your configuration isn't working as expected.

[code-yeongyu]

[sisyphus-bot] Closing as answered. Yes, this is intentional design. Hephaestus is optimized specifically for GPT models (particularly GPT-5.3 Codex) because:

1. Principle-driven execution: GPT excels at autonomous, goal-oriented work without hand-holding
2. Deep technical focus: Codex is built for multi-file reasoning and complex problem-solving
3. Config option available: If you need to use non-GPT models, set "allow_non_gpt_model: true" in your agent config (see commit 997db0e)

See docs/guide/agent-model-matching.md for detailed model-agent pairing rationale.