Skip to content
CC Switch WikiUnofficial Community Wiki

Configuration Not Applied or Lost After Switching Providers - cc-switch

Quick fix

Upgrade cc-switch to v3.16.2+ (v3.18.0 fixes nvm path detection). Put custom settings in “通用配置” to prevent overwrites; Linux save failures require v3.11.1+.

Symptom

Section titled “Symptom”
Error output
切换路由状态失败{{detail}}
English translation

Route status switching failed{{detail}}

Cause

Section titled “Cause”

This issue group covers configuration write defects in cc-switch across versions and platforms. In v3.15.0-v3.16.1, a failed route switch displayed the {{detail}} placeholder instead of a specific cause (#3727); on Windows, a mismatch between localhost and 127.0.0.1 bindings could cause route status switching to fail (#3727). In v3.16.0-v3.16.3, the GUI could not detect Node.js environments managed by nvm because PATH was too limited, preventing Codex/Claude Code upgrades or resulting in env: node: No such file or directory (#4448, #3584). Older versions rewrote entire configuration files when switching providers, causing custom plugin settings, MCP settings, or general configuration to be lost (#1459, #946, #2999). This was fixed in v4.0.0 by replacing only key fields. Linux had a bug that prevented saving edits to an individual provider (#1179). There was also a case where all provider configurations stopped working in the Codex desktop app after an upgrade (#4948).

Fix

Section titled “Fix”

  1. Upgrade to the latest version. The {{detail}} placeholder shown when route switching fails and the nvm detection issue were fixed in v3.16.2 and v3.18.0.

  2. If you use nvm to manage Node.js and see env: node: No such file or directory during an upgrade, as a temporary workaround run npm i -g @openai/codex@latest manually in a terminal, or create a symlink in a system path.

  3. If saving edits to an individual provider fails on Linux, upgrade to v3.11.1 or later.

  4. To prevent custom settings such as MCP and plugins from being overwritten when switching providers, put general settings in the “通用配置” section instead of an individual provider configuration, or upgrade to v4.0.0, which replaces only key fields such as the address, key, and model.

    ~/.claude/settings.json 或 Codex 对应配置文件
    // 在 cc-switch 的"通用配置"中填入自定义设置，而非依赖单个供应商配置

  5. If all configurations are lost after upgrading the Codex desktop app, try reverting to v3.16.4 or reconfigure in cc-switch and ensure the settings are written to “通用配置”.

Affected Versions

Section titled “Affected Versions”
Toolcc-switch
Version3.15.0 - 3.16.5 (路由/保存问题), 3.16.0 - 3.16.3 (nvm识别), 3.19.2 (Codex桌面端配置丢失)
PlatformsWindowsmacOSLinux

Source Issues

Section titled “Source Issues”

This page is distilled from 28 real issues

FAQ

Section titled “FAQ”
What should I do if Claude Code plugins or marketplaces disappear after switching providers?
Check whether there is a backup in cc-switch’s “通用配置”, or check the default provider imported after installation. From v4.0.0 onward, switching no longer overwrites the entire configuration, which can reduce this kind of issue.
Why does “切换路由状态失败” appear without a specific cause?
This was a bug in v3.16.1 and earlier: the {{detail}} placeholder in the error message was not replaced correctly. Upgrade to v3.16.2 or later to see the specific backend error details.
What should I do if cc-switch cannot detect Node installed with nvm or cannot upgrade Codex?
This is a known issue in v3.16.x (the GUI environment’s PATH is too limited). As a temporary workaround, run npm i -g @openai/codex@latest in a terminal to upgrade manually, or create a symlink for node in /usr/local/bin. This issue was fixed in v3.18.0.
Why do edits to an individual provider configuration have no effect after saving on Linux?
This was a bug in v3.11.0 and earlier. Upgrade to v3.11.1 or later.
Why do all provider configurations stop working after upgrading Codex?
A user reported this issue after upgrading to v3.16.5. Try reverting the version first, or put the settings in cc-switch’s “通用配置” section to prevent them from being reset.
How can I prevent cc-switch from overwriting my custom MCP settings?
In older versions, switching providers rewrote the entire configuration. Put custom MCP settings in “通用配置” (which applies to all providers) instead of an individual provider configuration, or upgrade to v4.0.0, which replaces only key fields and preserves other custom settings.
Why did editing general configuration save automatically?
Before v4.0.0, the general configuration editor dialog could save automatically under certain actions. This editing method was removed in v4.0.0; switching now replaces only key fields, while other settings remain in the original configuration file.
Section titled “Related problems”
Constant 429 errors: Claude Code AnyRouter errorConfirm the model mapping is blank and set "model": "opus[1m]"; if 429 errors continue, the issue may be with AnyRouter upstream, so try another network.1 issues mergednot installed or not executableUpgrade CC Switch to v3.16.2+, or set the CLI install directory by hand in settings, and clear noise printed by your shell at startup.4 issues mergedcc-switch Codex session history missingDisable Codex routing or unified history in cc-switch, make model_provider match old sessions, then restart Codex to restore history.7 issues merged

This is an unofficial community wiki with no affiliation to the cc-switch authors or the project itself. Its content is compiled from the project's public GitHub issues. This site distributes no software.