tinyhumansai/openhuman · error

Missing 'command' parameter

Error message

Missing 'command' parameter

What it means

ShellTool's execute_in_context extracts the required 'command' string before option parsing, timeout policy, and security classification run; this fires when args omit 'command' or pass a non-string. No process is spawned.

Source

Thrown at src/openhuman/tools/impl/system/shell.rs:293

        &self,
        args: serde_json::Value,
        _options: ToolCallOptions,
        context: Option<&ToolExecutionContext>,
    ) -> anyhow::Result<ToolResult> {
        self.execute_in_context(args, context).await
    }
}

impl ShellTool {
    async fn execute_in_context(
        &self,
        args: serde_json::Value,
        context: Option<&ToolExecutionContext>,
    ) -> anyhow::Result<ToolResult> {
        let command = args
            .get("command")
            .and_then(|v| v.as_str())
            .ok_or_else(|| anyhow::anyhow!("Missing 'command' parameter"))?;

        // Optional per-call wall-clock budget. `None`/`0` ⇒ run unbounded;
        // a positive value is clamped downstream by
        // `tool_timeout::explicit_call_timeout_*`. Shell has no default deadline
        // (issue #4023) — long scripts must run to completion.
        let requested_timeout = args.get("timeout_secs").and_then(|v| v.as_u64());

        let start = Instant::now();
        let (allowed, result) = self
            .run_with_security_in_context(command, requested_timeout, context)
            .await;
        let duration_ms = u64::try_from(start.elapsed().as_millis()).unwrap_or(u64::MAX);
        // `allowed` = passed the in-tool security checks. `approved` = the command
        // is Prompt-class (required human approval) and thus went through the
        // harness ApprovalGate to reach here — distinct from `allowed`. Reads and
        // Full-mode writes run without a prompt, so they audit as approved=false
        // rather than over-claiming a human approval that never happened. (The
        // gate's exact yes/no isn't threaded into tools; this is the accurate

View on GitHub (pinned to 7491200858)

Solutions

1. Include the "command" string to execute in the shell tool arguments

When it happens

Trigger: Thrown at src/openhuman/tools/impl/system/shell.rs:293 when the library encounters an invalid state.

Common situations: See trigger scenarios.

AI-assisted analysis of tinyhumansai/openhuman@7491200858 (2026-08-17). Data as JSON: /api/errors/a6a780342927a7e3. Report an issue: GitHub.