BigPizzaV3/CodexPlusPlus · error · anyhow::Error

工作目录不存在：{}

Error message

工作目录不存在：{}

What it means

The working directory passed to the Codex app-server, config.work_dir or the process cwd when empty, does not exist as a directory, so the connector aborts before spawning anything. The app-server would otherwise run in a bogus cwd and fail later in worse ways.

Source

Thrown at crates/codex-plus-core/src/connect/mod.rs:119

pub type SharedWeixinConnectStatus = Arc<Mutex<WeixinConnectStatus>>;

pub async fn run_weixin_connect(
    config: WeixinConnectConfig,
    stop: Arc<AtomicBool>,
    status: SharedWeixinConnectStatus,
) -> anyhow::Result<()> {
    let config = config.normalized();
    if config.token.is_empty() {
        bail!("请先扫码登录微信");
    }
    let work_dir = if config.work_dir.is_empty() {
        std::env::current_dir().context("无法读取当前工作目录")?
    } else {
        PathBuf::from(&config.work_dir)
    };
    if !work_dir.is_dir() {
        bail!("工作目录不存在：{}", work_dir.display());
    }

    update_status(&status, |current| {
        current.state = "starting".to_string();
        current.message = "正在连接微信和 Codex app-server...".to_string();
        current.account_id = config.account_id.clone();
        current.has_token = true;
    });

    let client = WeixinClient::new(&config.base_url, &config.token, &config.route_tag)?;
    let store = ConnectSessionStore::default_for_account(&config.account_id);
    let mut state = store.load().unwrap_or_default();
    let app_config = AppServerConfig {
        executable: config.codex_path.clone(),
        work_dir,
        model: config.model.clone(),
        sandbox: config.sandbox.clone(),
    };

View on GitHub (pinned to f2074595a2)

Solutions

1. Create the directory or point work_dir at an existing one
2. Store an absolute path and expand tilde or env vars before saving the profile
3. When work_dir is empty by design, start the process from a valid cwd
4. Fix typos in the saved profile path

Example fix

// before
run_weixin_connect(config, stop, status).await?;
// after: validate or create the directory before starting
let dir = std::path::PathBuf::from(&config.work_dir);
if !dir.is_dir() { std::fs::create_dir_all(&dir)?; }
run_weixin_connect(config, stop, status).await?;

let dir = std::path::Path::new(config.work_dir.trim());
if config.work_dir.trim().is_empty() || !dir.is_dir() {
    bail!("work dir missing: {}", dir.display());
}

When it happens

Trigger: WeixinConnectConfig.work_dir points at a deleted or renamed directory; a relative path resolved against a different cwd; an unexpanded tilde or env var left in the saved path; empty work_dir while the daemon cwd itself was removed.

Common situations: Project moved or deleted after the profile was saved; service started from a different directory; the volume holding the directory is not mounted.

Related errors

• 请先扫码登录微信
• 微信连接已停止
• 获取微信登录二维码失败：HTTP {status}
• 微信二维码响应缺少必要字段
• 查询微信扫码状态失败：HTTP {status}

AI-assisted analysis of BigPizzaV3/CodexPlusPlus@f2074595a2 (2026-08-23). Data as JSON: /api/errors/e99bba071c9ab538. Report an issue: GitHub.