Add configurable verbosity for OpenAI Responses API

[thnk2wn]

Add configurable verbosity for OpenAI Responses API

Fixes #2775

Problem

Models like gpt-5.2-codex and other newer OpenAI models only support medium reasoning and verbosity levels, but the codebase was using low by default. This caused 400 Bad Request errors:

Failed to stream openai-responses chat: openai 400 Bad Request:
Unsupported value: 'low' is not supported with the 'gpt-5.2-codex' model.
Supported values are: 'medium'.

Solution

This PR implements a scalable, user-configurable approach instead of hardcoding model-specific constraints:

1. Changed default verbosity from "low" to "medium" - more widely supported across OpenAI models
2. Added ai:verbosity config option - allows users to configure verbosity per model in waveai.json
3. Changed rate limit fallback from low to medium thinking level for better compatibility
4. Removed hardcoded model checks - solution is scalable for future models

Changes

Backend Changes

• pkg/aiusechat/openai/openai-convertmessage.go - Use configurable verbosity with safe defaults
• pkg/aiusechat/uctypes/uctypes.go - Add Verbosity field to AIOptsType
• pkg/aiusechat/usechat.go - Pass verbosity from config to options
• pkg/wconfig/settingsconfig.go - Add Verbosity to AIModeConfigType

Schema Changes

• schema/waveai.json - Add ai:verbosity with enum values (low/medium/high)
• frontend/types/gotypes.d.ts - Auto-generated TypeScript types

Configuration Example

Users can now configure both thinking level and verbosity per model:

{
  "openai-gpt52-codex": {
    "display:name": "GPT-5.2 Codex",
    "ai:provider": "openai",
    "ai:model": "gpt-5.2-codex",
    "ai:thinkinglevel": "medium",
    "ai:verbosity": "medium"
  }
}

[CLAassistant]

All committers have signed the CLA.

[coderabbitai]

Walkthrough

This pull request adds an optional AI verbosity configuration field across the codebase. It introduces "ai:verbosity" (enum "low", "medium", "high") to TypeScript, Go structs, and the JSON schema; adds Verbosity to AIOptsType; sets a default OpenAIDefaultVerbosity = "medium" and wires verbosity into OpenAI request construction; and adjusts premium rate-limit thinking level handling to use medium instead of low.

Estimated code review effort

🎯 2 (Simple) | ⏱️ ~12 minutes

🚥 Pre-merge checks | ✅ 4 | ❌ 1

❌ Failed checks (1 warning)

Check name	Status	Explanation	Resolution
Docstring Coverage	⚠️ Warning	Docstring coverage is 50.00% which is insufficient. The required threshold is 80.00%.	Write docstrings for the functions missing them to satisfy the coverage threshold.

✅ Passed checks (4 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title accurately summarizes the main change: adding configurable verbosity for OpenAI Responses API, which is the primary objective of this PR.
Description check	✅ Passed	The description is detailed and directly related to the changeset, explaining the problem, solution, and providing configuration examples that match the code changes.
Linked Issues check	✅ Passed	The PR successfully addresses all coding requirements from issue #2775: changed default verbosity to 'medium', added configurable ai:verbosity option, updated rate-limit fallback, and removed hardcoded model checks.
Out of Scope Changes check	✅ Passed	All changes are directly scoped to implementing configurable verbosity for OpenAI Responses API; no unrelated modifications or out-of-scope additions are present.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

✨ Finishing touches

• 📝 Generate docstrings

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands and usage tips.}