What causes "hive agent '{}' exited with status {:?}" in nikivdev/code?

Check the hive output printed above the error (stdio is inherited) for the real cause

hive agent ' ' exited with status

Error message

hive agent '{}' exited with status {:?}

What it means

`run_agent` spawns `hive agent <agent> <prompt>` with inherited stdio and waits via `.status()`. If the hive process exits with a failure code, the function bails with this message naming the agent and the exit status (as `Option<i32>` via `{:?}`). The actual reason for failure must come from hive's own output, which was streamed to the terminal.

Solutions

Check the hive output printed above the error (stdio is inherited) for the real cause
Verify the agent name exists: `hive agent <name>` or list available agents
Run the exact command manually to reproduce: `hive agent <agent> "<prompt>"`
If the status shows `None`, the process died by signal — check dmesg/limits (OOM) or rerun

Example fix

// before
hive::run_agent("reviwer", "do the thing")?; // typo'd agent
// after
let agent = "reviewer"; // exact, existing agent name
match hive::run_agent(agent, "do the thing") {
    Err(e) => { eprintln!("agent {agent} failed: {e}"); }
    ok => ok?,
}

Defensive patterns

Strategy: try-catch

Validate before calling

let known = ["reviewer", "planner", "deploy-agent"];
anyhow::ensure!(
    known.contains(&agent),
    "unknown hive agent '{}' — check `hive agent --list`",
    agent
);
anyhow::ensure!(hive_available(), "hive not on PATH");

Type guard

null

Try / catch

match hive::run_agent(agent, prompt) {
    Err(e) if e.to_string().contains("exited with status") => {
        eprintln!("hive agent failed; its output was streamed above — inspect it");
        if e.to_string().contains("None") {
            eprintln!("process died by signal (killed/OOM/Ctrl-C)");
        }
    }
    r => r?,
}

Prevention

Validate agent names against hive's agent list before invoking
Capture or tee hive output to a log file since the error only carries the exit code
Set retry/backoff for transient hive backend failures
Monitor for signal exits (status None) which indicate kills, not hive errors

When it happens

Trigger: hive exiting non-zero for an unknown agent name, invalid prompt, its own internal error, or being killed by a signal (in which case `status.code()` is `None`, printed as `None`).

Common situations: Passing an agent name that has no corresponding hive agent definition; hive failing authentication or hitting an API rate limit; prompt rejected by hive; process killed by OOM-killer or Ctrl-C resulting in a signal exit.

Related errors

AI-assisted analysis of nikivdev/code@a747e741ae (2026-09-01). Data as JSON: /api/errors/3ef5b4c1db624d65. Report an issue: GitHub.

Appendix: source

Thrown at src/hive.rs:325

/// Run a hive agent with a prompt
pub fn run_agent(agent: &str, prompt: &str) -> Result<()> {
    // Check if hive is available
    if which::which("hive").is_err() {
        anyhow::bail!("hive not found on PATH. Install from https://github.com/example/hive");
    }

    let status = Command::new("hive")
        .arg("agent")
        .arg(agent)
        .arg(prompt)
        .stdin(Stdio::inherit())
        .stdout(Stdio::inherit())
        .stderr(Stdio::inherit())
        .status()
        .context("Failed to run hive")?;

    if !status.success() {
        anyhow::bail!(
            "hive agent '{}' exited with status {:?}",
            agent,
            status.code()
        );
    }

    Ok(())
}

/// Run an agent interactively (prompt via stdin)
pub fn run_agent_interactive(agent: &str) -> Result<()> {
    if which::which("hive").is_err() {
        anyhow::bail!("hive not found on PATH");
    }

    let status = Command::new("hive")
        .arg("agent")
        .arg(agent)

View on GitHub (pinned to a747e741ae)