heyhuynhgiabuu/proxypal v0.4.24

v0.4.24 — Qwen OpenCode config + CLIProxyAPIPlus v6.9.22-0

on GitHub

What's New in v0.4.24

🔧 Sidecar Upgrade: CLIProxyAPIPlus v6.9.22-0

Upgraded sidecar from v6.9.19-0 → v6.9.22-0.

Key sidecar changes:

• fix(executor): Immediate retry with token refresh on 429 for Qwen
• fix(executor): OAuth tool name remapping with rename detection
• fix(claude): Remap OAuth tool names to Claude Code style (avoids third-party fingerprint detection)
• fix(claude): Preserve OAuth tool renames when filtering tools
• feat(antigravity): Configurable signature cache with bypass-mode validation
• feat(antigravity): Prefer prod URL as first priority
• fix(handlers): Bind listener on all interfaces instead of localhost only

✨ Qwen Full Model Config for OpenCode

When configuring OpenCode via the Configure button, Qwen models now get a complete model spec:

{
  "attachment": true,
  "reasoning": true,
  "temperature": true,
  "tool_call": true,
  "modalities": {
    "input": ["text", "image"],
    "output": ["text"]
  },
  "options": {
    "reasoningEffort": "<user-configured>",
    "reasoningSummary": "auto",
    "textVerbosity": "medium"
  },
  "variants": {
    "xhigh": { "include": ["reasoning.encrypted_content"], "reasoningEffort": "xhigh", "reasoningSummary": "auto" },
    "high":  { "include": ["reasoning.encrypted_content"], "reasoningEffort": "high",  "reasoningSummary": "auto" },
    "medium": { "reasoningEffort": "medium", "reasoningSummary": "auto" },
    "low":    { "reasoningEffort": "low",    "reasoningSummary": "auto" }
  }
}

This enables image attachments, reasoning variants (xhigh/high/medium/low), and proper tool call support for all Qwen models in OpenCode.