Why should you care?

Two agents updating the same ERP journal need coordination tokens — otherwise duplicate postings appear.

See it live — copy this example

Paste into an ASP.NET Core 8+ / AgentNest project, then run with dotnet run (set your API keys in user-secrets).

// AgentNest.Coordination/ResourceLockService.cs
public sealed class ResourceLockService(IDistributedCache cache)
{
    public async Task<IAsyncDisposable?> AcquireAsync(string resourceKey, TimeSpan ttl, CancellationToken ct)
    {
        var token = Guid.NewGuid().ToString("N");
        await cache.SetStringAsync($"lock:{resourceKey}", token,
            new DistributedCacheEntryOptions { AbsoluteExpirationRelativeToNow = ttl }, ct);
        return new Releaser(cache, resourceKey, token);
    }
}

What happened?

• ResourceLockService grants short TTL locks on keys like erp:journal:2025Q1.
• Agents await AcquireAsync before write tools run.

Practice next

1. Wrap write kernel functions with lock acquisition.
2. Use RedLock.net for production Redis clusters.
3. Return 409 Conflict when lock unavailable.
4. Implement lock renewal heartbeat for long ERP batches.
5. Shard lock keys by tenantId prefix.

Remember

Coordinate agents with distributed locks on business keys. Always set TTL on locks. Return explicit conflicts instead of silent overwrites.

Duplicate journal prevention

Two ERP agents retry the same month-end posting after timeout.

Outcome: ResourceLockService ensures only one journal draft commits.