Show HN: Agentic Task Queue – batch processing of agentic tasks • Hacker News Board

Show HN: Agentic Task Queue – batch processing of agentic tasks

#1

Post by endyai » Tue, Mar 24, 2026, 7:39 PM UTC

I had a huge number of database rows that each needed LLM reasoning and tool use.

Existing harnesses like Claude Code don't handle this well - give it a batch and it'll try to write you a script instead of actually reasoning over each item.

Even when it works, context bloats after each turn - gets expensive, slow, and starts hallucinating. And there's no built-in retry handling or parallelization.

So I built this tool. Define the task as a prompt, pipe in your task list, set concurrency, model, tools, etc.

Open to feedback!

Re: Show HN: Agentic Task Queue – batch processing of agentic tasks

#2

Post by ngarner » Wed, Mar 25, 2026, 1:26 AM UTC

The lack of built-in retries is a huge pain, but the bigger risk is a "successful" retry that just outputs another hallucination.

How are you defining a "success" signal for these tasks? Is it just a 200 OK, or are you planning a fidelity audit for each item in the queue to trigger those retries?

Re: Show HN: Agentic Task Queue – batch processing of agentic tasks

#3

Post by endyai » Wed, Mar 25, 2026, 2:09 PM UTC

ngarner wrote: post #2

The lack of built-in retries is a huge pain, but the bigger risk is a "successful" retry that just outputs another hallucination. How are you defining a "success" signal for these tasks? Is it just a 200 OK, or are you planning a fidelity audit for each item in the queue to trigger those retries?

You can handle this in a few ways depending on the task. Even. adding to the prompt "double check your answer before answering" - the agent will take another turn to double check its work. You can also do this with a fresh task/prompt.

Ideally, if you are able to use code to validate (either with a test or eval) that works best.