zeroclaw-labs/zeroclaw · error

domain '{}' is not in browser_delegate allowed_domains

Error message

domain '{}' is not in browser_delegate allowed_domains

What it means

When browser_delegate's allowed_domains is non-empty it acts as a strict allowlist: the URL's domain must match an entry exactly or as a subdomain (domain_matches: equal, or ends with ".pattern"), otherwise the task is rejected. An empty allowed_domains list means no restriction beyond the denylist.

Source

Thrown at crates/zeroclaw-tools/src/browser_delegate.rs:114

            anyhow::bail!("URL has no host: {}", url);
        }

        // Check blocked domains first (deny takes precedence)
        for blocked in &self.config.blocked_domains {
            if domain_matches(&domain, blocked) {
                anyhow::bail!("domain '{}' is blocked by browser_delegate policy", domain);
            }
        }

        // If allowed_domains is non-empty, it acts as an allowlist
        if !self.config.allowed_domains.is_empty() {
            let allowed = self
                .config
                .allowed_domains
                .iter()
                .any(|d| domain_matches(&domain, d));
            if !allowed {
                anyhow::bail!(
                    "domain '{}' is not in browser_delegate allowed_domains",
                    domain
                );
            }
        }

        Ok(())
    }
}

/// Check whether `domain` matches a pattern (exact or suffix match).
fn domain_matches(domain: &str, pattern: &str) -> bool {
    let d = domain.to_lowercase();
    let p = pattern.to_lowercase();
    d == p || d.ends_with(&format!(".{}", p))
}

/// Maximum stderr bytes to capture from the subprocess.

View on GitHub (pinned to 88bb9c8533)

Solutions

1. Add the exact domain (or its parent) to browser_delegate allowed_domains in config
2. Verify the matching semantics: suffix matching covers subdomains only, never sibling TLDs
3. Keep the allowlist maintained alongside the services the agent is authorized to touch

Example fix

# config.toml — before
[browser_delegate]
allowed_domains = ["example.com"]

# after
allowed_domains = ["example.com", "api.example.org"]

fn domain_matches(domain: &str, pattern: &str) -> bool {
    let (d, p) = (domain.to_lowercase(), pattern.to_lowercase());
    d == p || d.ends_with(&format!(".{p}"))
}
let allowed = allowed_domains.iter().any(|p| domain_matches(&host, p));
if !allowed_domains.is_empty() && !allowed {
    return Err(format!("host '{host}' not allowlisted"));
}

match delegate.execute(args).await {
    Ok(res) if res.success => { /* ... */ }
    Ok(res) => {
        if res.error.as_deref().unwrap_or_default().contains("not in browser_delegate allowed_domains") {
            // add the domain to config or drop the URL from the task
        }
    }
    Err(e) => return Err(e),
}

When it happens

Trigger: A task referencing https://api.example.org while allowed_domains=["example.com"] — example.org does not match; sibling domains and different TLDs never match a parent entry.

Common situations: Forgetting the allowlist is exact-or-child-domain: docs.example.com passes with entry example.com, but example.co.uk does not; new microservices on separate domains need config updates; typos in either the URL or the config entry.

Related errors

• domain '{}' is blocked by browser_delegate policy
• Host '{host}' is not in browser.allowed_domains
• Workspace path {} is not in runtime.docker.allowed_workspace
• Serial path not allowed: {}. Allowed: {}
• memory write blocked by content scan: {kinds}

AI-assisted analysis of zeroclaw-labs/zeroclaw@88bb9c8533 (2026-08-23). Data as JSON: /api/errors/75329a756f1bbab6. Report an issue: GitHub.