n8n’s Post

Right now Ask Steve (https://www.asksteve.to) calls an LLM in the background with your prompt and returns the result. But really it's just calling an API which happens to be an LLM. With a few small new features in the latest version of Ask Steve, I've made it easy to call ANY API. So now rather than triggering a single LLM call from the current page or selection, you can do... anything! Pre-process the input, augment it with external information, log it, etc. Then process the input with code, external services, one or more LLMs, or some combination. And finally post-process the result before returning it to the user in the Ask Steve UI. To demo this, I created a simple workflow using n8n.io - a really powerful low/no-code workflow automation builder - and connected it to an Ask Steve Skill. This workflow takes the selected text, parses it, puts it into a Google Sheet and returns a confirmation. You can do this with your own code running on a server somewhere, or with any visual workflow builder like n8n, Gumloop (YC W24), Activepieces (YC S22), Make, Zapier etc. that supports Webhooks. Ask Steve has leveled up :)

Simple and powerful. Check out S3’s new conditional write feature & how to use it. — 1. Naming Strategy Use a structured naming convention for objects, including the version number as part of the object key. For example, an order might be stored as ecommerce/tenantA/orders/12345/001.json. The `001` in the key represents the order's version. Every time the order is updated, a new object with an incremented version number (e.g., 12345:002.json) is created. 2. Using `If-None-Match` for Conditional Writes: Use the If-None-Match=”*” header when updating an order. This tells S3 to only proceed with the write operation if an object with the specified key (e.g. order123:002.json) does not exist. This prevents two processes from creating the same version number, ensuring each update creates a new, unique version. 3. Handling Conflicts: If the If-None-Match=”*” condition fails (resulting in a `412 Precondition Failed` response), it means another process has already created that version. The process can then reread the latest version, resolve the conflict, and retry the operation with the next version number.

Here is a quick way to detect default Metasploit behavior. If the threat actor did not change their executable template for their payload, the OriginalFileName will be "ab.exe" and the Company will be "Apache Software Foundation". If the OriginalFileName != Image you could do some extra filtering to scrutinize what is going on. Below is a screenshot where I created 2 new columns making the data inside all lower case and split out the filename from the Image field. I am comparing to see if the OriginalFileName is in the Image column. From here you could pivot on the ProcessGuid field to see what else these processes have been up to.

I find it helpful to queue up things I want to read for a closer look the following day (delayed gratification, and avoiding constantly refreshing the same sources). pwr is how I implement this terminal-centric workflow. Maybe it's useful to you as well?

Auto-instrumentation is great, but it shouldn't be your goal post. Adding custom spans demystifies the application's behaviour and choices. More importantly, it increases the accessibility and debugging of your production system. I've encountered many traces where there's a span representing the service followed by a sea of heterogeneous spans. This doesn't communicate much and is especially hostile to the unfamiliar. For starters, you need to examine the details of each span and understand what the operation is doing first. For some spans that means reading the database queries and trying to understand what it's doing. Often arguments are unavailable which completely obscures the intent. Then, after understanding what each span represents, you now need to speculate about why. It provides more questions than answers. Consider the trace illustrated below. - Why are we selecting a bunch of records, inserting into one table, and then updating another? - Why are we pulling data from the cache? Was the fetch from the cache successful? - What's the expected system behaviour for a cache hit/miss? What should you do instead? Add spans that illuminate internal system behaviour. One successful approach we've used internally is adding spans for each component method. Within an application we'll have an API component call a service component, which calls a repository component, or external client, and so on. Adding spans for each of these component methods bakes intentionality and explains what the system is doing. Instead of a sea of spans, you'd see a `EventService.get_events` span, or `EventStateService.update_state` span. Each of those spans would contain tags that capture artifacts from the method. It gives a much clearer picture of what the system is doing. #observability

TLDR: I wrote a guide on how to handle 3rd party rate limitations with a queue comparing the best tools of 2023.  https://lnkd.in/edMSSDUP 3rd party rate limitations require us to build workarounds in order to, eventually, fulfill some tasks. Among those workarounds is using a queue, but even with a queue, we still need to implement some client logic. Here is a short guide on how to do so using queue methods advantages.

today I saw an error message that simply stated the workflow has failed because there is no error. This sentence is both funny and confusing. Do you know that if there was an error the wf wouldn't have failed? What happens behind the scene is poor error handling. In this scenario a process trigger an async task, but if the async task fails with weak error handling, the task that initially started the async task has no way of knowing what really happened. The programers usually hard code in their catch block a string that starts the beginning of the error message appended by the error message returned. So the workflow has failed is a fixed string thats is always return. And here "there is no error" simply states that the async task failed but the failure is not handled property to return a proper reason. If you've made it this far, thanks for reading. Please like and let me know in the comment if you'd like to share more stories like this.