What causes "failed to resume codex session {}" in nikivdev/code?

Verify the session ID exists in the codex session/runtime state listing

failed to resume codex session {}

Error message

failed to resume codex session {}

What it means

After computing a resume plan, the tool attempts to resume the existing codex session via the resume bridge. If the bridge reports that no session was actually resumed/launched, it bails with the failing session ID. It signals the stored session could not be attached to.

Source

Thrown at src/ai.rs:11749

                .session_id
                .as_deref()
                .ok_or_else(|| anyhow::anyhow!("missing session id for resume plan"))?;
            println!(
                "Opening Codex session {} in {}...",
                truncate_recover_id(session_id),
                launch_path.display()
            );
            if launch_session_for_target(
                session_id,
                Provider::Codex,
                plan.prompt.as_deref(),
                Some(&launch_path),
                plan.runtime_state_path.as_deref(),
                plan.trace.as_ref(),
            )? {
                Ok(())
            } else {
                bail!("failed to resume codex session {}", session_id);
            }
        }
        "new" | "recover-new" => {
            maybe_open_cursor_for_pr_feedback_check(plan);
            new_session_for_target(
                Provider::Codex,
                plan.prompt.as_deref(),
                Some(&launch_path),
                plan.runtime_state_path.as_deref(),
                plan.trace.as_ref(),
            )
        }
        other => bail!("unsupported codex open action: {}", other),
    }
}

fn maybe_open_cursor_for_pr_feedback_check(plan: &CodexOpenPlan) {
    let Some(query) = plan

View on GitHub (pinned to a747e741ae)

Solutions

Verify the session ID exists in the codex session/runtime state listing
Start a fresh session (`new` action) if the old rollout is gone
Check codex CLI version compatibility and re-login/repair the runtime

Defensive patterns

Strategy: try-catch

Validate before calling

// confirm the session/runtime state exists before resuming
let states = codex_runtime::load_runtime_states()?;
if !states.iter().any(|s| s.session_id == session_id) {
    eprintln!("{} has no runtime state; use new session", session_id);
}

Type guard

fn resumable(states: &[RuntimeState], id: &str) -> bool {
    states.iter().any(|s| s.session_id == id)
}

Try / catch

match result {
    Err(e) if e.to_string().starts_with("failed to resume codex session") => new_session_for_target(Provider::Codex, ...),
    other => other,
}

Prevention

Don't prune ~/.codex session/rollout files while sessions are active
Keep codex CLI version consistent between machines
Treat resume failure as a signal to start a new session

When it happens

Trigger: `f codex open <session>` with action resume/recover where `run_codex_resume...` returns false — e.g. the runtime state or rollout file for that session ID is missing, corrupted, or the underlying process could not be launched.

Common situations: Resuming a session from an old machine after upgrading codex (rollout format/path changed); deleted `~/.codex/sessions` history; passing an ID copied from logs that was never persisted.

Related errors

AI-assisted analysis of nikivdev/code@a747e741ae (2026-09-01). Data as JSON: /api/errors/3645c145e095041b. Report an issue: GitHub.