deniz.in

Markets

Weather

Loading weather

· via Hacker News – Front Page (native)

ADE Index ranks 975 agentic coding tools with sourced facts and published weights

ADE Index, launched on Hacker News, ranks 975 agentic coding tools across 383 agent workspaces, with every fact tied to a cited source and a published scoring methodology.

ADE Index ranks 975 agentic coding tools with sourced facts and published weights

A new project called ADE Index launched on Hacker News on 19 November 2026 via a Show HN post, presenting a ranked, source-annotated catalogue of agentic coding tools. The post's title counts 975 tools, while the site's own methodology page says the listings currently number 999 — a gap that suggests the catalogue is still growing. At the centre are the 383 tools the site classifies as agent workspaces: environments built to run and supervise coding agents.

What the index contains

Agent workspaces are grouped by how you use them. The largest sections cover desktop apps (240 listings), browser-based and self-hosted interfaces (111), terminal and TUI tools (57), and remote or mobile supervision tools (81), with seven further categories elsewhere on the site. Tools can appear in several sections at once, so the counts overlap.

According to the site, only 109 of the 383 workspaces have an index score so far, with the first rankings published on 18 November 2026.

The current leaderboard mixes familiar names with newer entrants. Herdr, a terminal workspace, leads with a score of 66, ahead of Paseo (53), DeepSeek Harness (52), T3 Code from Ping.gg (47), cmux (42), ZCode (39), Warp (34) and Goose (33), with OpenHands and Obeya tied at 27. Each listing carries structured details the index says it gathered from cited sources: Paseo is Apache-2.0 licensed with a free tier and about 24,300 GitHub stars; T3 Code is MIT licensed at roughly 64,200 stars and adds phone-based approval of agent actions; cmux is source-available under GPL-3.0-or-later with a BUSL-1.1 server component, priced at $40 per month billed yearly or $50 monthly; OpenHands is MIT licensed at about 93,700 stars, with the listing recording a $500 per month price point.

The index also reads GitHub release feeds, and it counted 169 agent workspace releases shipped in the past week alone — a measure of how quickly the category is moving.

A comparison view lets you place up to four tools side by side, and the configuration lives in the URL, so a set-up comparison can be shared as a link.

How the score is built

The methodology page publishes the weights. The base of the score is a tool's traction on Hacker News, applied identically to open-source and closed-source products. GitHub stars are displayed alongside the score but deliberately not folded into it. Community votes can only ever raise a score, contributing at most 20% of it, graded on a square-root curve measured against the most-voted tool and with a floor of 100 effective votes. Votes decay with a 90-day half-life, makers cannot vote for their own products, and vote counts below ten are hidden for display purposes only.

Sourcing, maker claims and sponsors

The site says 21,453 sources stand behind its listings, and any dotted-underlined fact on a tool page can be opened to reveal the sentence the indexers read, where they read it and when. Tool makers can claim their listing to keep the description, screenshots, links and pricing current, and those details are marked as maker-supplied — but scores, ranks, votes and quoted material stay out of the vendor's reach. Sponsored placements are labelled with who paid, never appear as rows inside a ranked list, and the ranking pages themselves carry none. The site also states that it runs no affiliate links.

Why it matters

The agentic coding tool market has become genuinely hard to navigate: hundreds of workspaces, weekly launches, and marketing claims that rarely survive contact with a README. An index that ties every factual claim to a source sentence, keeps GitHub stars out of the ranking, caps the influence of votes and publishes its weights gives engineering teams a defensible starting point for evaluation — and a template for how tool directories could disclose their methods. The caveats are real: a score rooted in Hacker News attention will favour tools that resonate with that audience, only 109 of 383 workspaces are scored so far, and the vote component remains a popularity signal even at a 20% cap. On transparency alone, though, this sets a higher bar than most AI tool lists currently circulating.

  • #agentic-coding
  • #developer-tools
  • #ai-agents
  • #hacker-news
  • #tool-directory

Related posts