Hmbown/CodeWhale · error

Selected Fleet is missing or unreadable; inspect /fleet and…

Error message

Selected Fleet is missing or unreadable; inspect /fleet and repair or clear the selection.

What it means

Thrown when resolving the user's selected Fleet fails because the Fleet store cannot be read (`resolve_selected_fleet` returns Err). This typically means the selection file exists but is corrupt, unreadable, or points at something broken. The message directs the user to the /fleet command to repair or clear the selection.

Solutions

1. Run /fleet in the TUI and clear the stale selection, then re-select a valid Fleet.
2. Inspect the fleet store file for corruption or schema drift and fix or delete it.
3. Check file permissions on the store location so the current user can read it.
4. As a workaround, pass an explicit route override for the session, which bypasses Fleet resolution.

match start_turn(...).await {
    Err(e) if e.to_string().contains("Selected Fleet is missing or unreadable") => {
        clear_fleet_selection();
        retry_without_fleet();
    }
    other => other,
}

When it happens

Trigger: Starting a turn with a persisted Fleet selection whose store file cannot be loaded: corrupted JSON, bad permissions, or an IO error reading the selection. Skipped entirely when an explicit route override is active.

Common situations: Manually editing the fleet selection file and breaking its format; upgrading versions changed the store schema; file permissions changed after switching users; disk full during a previous write.

Understand the failure class

Background: "Config file not found": what it means and how to fix it in docker-sync, Maven, Vagrant, Turborepo and other tools — this error's family across 60 libraries.

Related errors

• agent profile path is not a directory
• agent profile provider cannot be empty
• agent profile provider must be a simple provider id
• agent profile tools.posture= would widen permissions; use…
• cannot create Fleet run

AI-assisted analysis of Hmbown/CodeWhale@73e0f67d83 (2026-09-22). Data as JSON: /api/errors/d157eb229cd5cf81. Report an issue: GitHub.

Appendix: source

Thrown at crates/tui/src/lib.rs:8302

/// Apply the selected v2 Fleet's operator to a fresh root session.
///
/// Provider/model are one atomic route: any explicit launch override for
/// either half keeps the caller's full route and bypasses the saved operator.
/// Reasoning is independent, so an explicit reasoning flag keeps its value
/// while the Fleet may still select the provider/model pair. Resumes call
/// this helper only for fresh sessions and therefore retain their saved route.
fn apply_selected_fleet_operator_for_launch(
    config: &mut Config,
    workspace: &Path,
    explicit_route_override: bool,
    explicit_reasoning_override: bool,
) -> Result<bool> {
    if explicit_route_override {
        return Ok(false);
    }
    let Some(selected) = crate::fleet::store::resolve_selected_fleet(workspace).map_err(|_| {
        anyhow!(
            "Selected Fleet is missing or unreadable; inspect /fleet and repair or clear the selection."
        )
    })?
    else {
        return Ok(false);
    };
    let fleet_name = crate::safe_label::SafeLabel::phrase(&selected.name);
    let (fleet, _) = crate::fleet::store::load_fleet_at(&selected.path).map_err(|_| {
        anyhow!(
            "selected Fleet '{}' ({}) is invalid or unreadable; inspect /fleet and repair or clear the selection.",
            fleet_name,
            selected.scope.label()
        )
    })?;
    let Some(operator) = fleet.operator.as_ref() else {
        return Ok(false);
    };
    let provider_id = operator.provider.trim();

View on GitHub (pinned to 73e0f67d83)