Settings

Theme

Show HN: ADE Index – a sourced and ranked index of 975 agentic coding tools

adeindex.com

25 points by jonnyasmar · 10 comments

Reader

3 threads
rabidheuristics

This site is… Not great. Go look for instance at Pi, and its comparison to Oh-My-Pi. The comparisons themselves are data points that don’t actually say much about the product, don’t actually describe their capabilities, and (critically) don’t tell me what is better or worse between one or the other.

This looks like it was built in a short amount of time through a set of agents, with no real thought put into the reason you would care about a site like this. Sentiment analysis is cool I guess, but pretty worthless for someone actually looking for a new agent harness or tool to try.

With the harness becoming just as important (If not more important) than the model itself this could have a ton of value- but as it stands I hope the hosting isn’t costing you a whole lot.

  • jonnyasmarOP

    This is absolutely fair critique. It was indeed built very fast. Much of the initial work was focused on indexing and setting up the automated pipelines. Almost 1000 tools is a significant number to have agents scour, consume, and parse from very disparately shaped sources (including the various sentiment sources I'm looking at).

    The idea behind the sentiment is to help act as a pre-filter or social proof and allow further drill in from there because no one wants to filter through 1000 tools by hand when their capabilities can be technically highly similar.

    I do actually have routines set up, however, to regularly assess substance and content quality and the sentiment allows my agents to learn which ones to prioritize.

    Your feedback here is absolutely heard and noted. Trust it will be incorporated! Thanks!

danpalmer

Checks Antigravity

> Not ranked yet: no Hacker News mentions in the last 90 days

> Capabilities 2/8 – Read by ADE Index in the tool’s own docs. A capability not shown is not established yet; it may still be there.

> Antigravity IDE's screenshot is actually of the Antigravity App.

This site is slop. Very bad data quality, and the data is literally the whole point.

ajasmin

How many of these things do we really need. Especially given a lot of what's described as agentic engineering seems comes down to writing a few markdown files...

  • codedrift_

    Right now, I think there's going to be a groundswell of these tools. We don't know what's "right" or "best" yet. I'm getting 2000x flashbacks, where it was the wild west for web technologies. I'm more shocked there's only a thousand or so listed so far. Submissions are down or I'd have tossed a few more into there. Give it about 6-8 months and we'll probably see common themes emerge.

    One early theme I've noticed is ADEs are getting smarter about answering if they're blocked. An idle chat that ended in a question mark is probably blocked because it's waiting on human input. But it also looks identical to an idle thread that finished its work.

    We'll probably need something like these ADEs though, even if it's just VSCode with an Agent View. You outgrow markdown files quickly. At some point you have to enforce the markdown files, make sure workflows are adhered to, etc. At that point, you start looking for a tool.

    • jonnyasmarOP

      Oh.... submissions should definitely not be down. Appreciate you trying to add! Will get that fixed ASAP.

      Definitely agree with a lot of this. I've come to learn after building an ADE for 8 months that the most important piece of it is how to let the human know when & where they're needed. As you start parallelizing a ton of work, your attention because exceedingly precious.

      One of the techniques my ADE uses is that it encourages agents to use decision forms instead of plain language questions or they're driven by tasks (which can be 100% agentically managed). This let's me keep track of asks from directors (aka orchestrators/bots/long-lived agents) or short-lived sessions and surface them efficiently.

      AGENTS.md and skills are valuable, but they're still non-deterministic. Efficiency and trust in a tool comes from deterministic guarantees. The second you have to comb through all of your work to see where you need to follow-up or nudge an agent along, all of the efficiencies of AI start to evaporate.

  • jonnyasmarOP

    As someone who built one and went into it in a crowded market 8 months ago when there were like 10, I thought it was going to be a lot. I knew the number had grown significantly, but not 100x

    There's a lot of awesome stuff, but also a lot of garbage.

    That said, some of these tools are proof that it's really way more than just writing a few markdown files. I did that whole thing for a while, md + iTerm and I was good. But having a robust ADE is absolutely game-changing.

Keyboard Shortcuts

j
Next item
k
Previous item
o / Enter
Open selected item
?
Show this help
Esc
Close modal / clear selection