Radix

Why STEADY

STEADY (70), at the base of the band, because the concept is coherent and genuinely agent-native and the demand signal is strong (~412 Hacker News points), but the public surface Hlido can actually verify is minimal — a spare landing page, an install command and a key-gated signup for a macOS app — so the broad artifact capabilities are asserted, not demonstrated. Confidence is low-medium accordingly: not lower, because nothing on the surface reads as dishonest or broken; not higher, because there is little to test without installing the app.

Public-surface checklist

• PASS Homepage loads (required) — radix-os.com returned a landing page (page_title 'Radix — a visual UI for agentic programming')
• PASS Primary value prop (required) — 'A visual UI for agentic programming — build persistent flexible artifacts as your agent works'
• PASS Cta present (required) — 'GET A FREE KEY' / 'GET A KEY' and an install command
• PASS Pricing or access — A free key on signup; a pricing page is linked in the nav but specific tiers were not present on the captured landing
• FAIL Evidence or demo — The landing shows only a one-line pitch, an install command and a word-cloud of artifact types; no worked example, demo or documentation of the artifacts was present on the captured surface

What it does well

• A genuinely agent-native premise — persistent, inspectable artifacts a coding agent builds and keeps wired to the repo, rather than throwaway chat output
• Strong, timely demand signal (~412 points on Hacker News at launch) for a real developer pain point
• A hinted breadth of artifact types (flame graphs, stack traces, deadlock/false-sharing analysis, N+1 detection, schemas, state machines, import cycles, rate limiters) that maps directly onto real debugging and design work
• A clean, low-friction entry path — a single install command and a free key
• A focused, uncluttered surface that states its one idea plainly

What it fails at

• The public landing is deliberately minimal — a one-line pitch, an install command and a word-cloud — leaving the actual capabilities undescribed and unverifiable from the web
• The product is a native macOS app behind a signup key, so its behaviour cannot be tested from the public surface (and macOS-only narrows reach)
• No public repository or documentation surface was found to corroborate the artifact-building claims
• The broad list of artifact types is presented as a teaser, without worked examples or evidence of each on the public surface
• No hands-on run by Hlido — the persistence, the repo-wiring and the artifact quality are asserted, not confirmed here

Pricing & access

• Pricing findable on the public surfacePASS A free key on signup; a pricing page is linked in the nav but specific tiers were not present on the captured landing (tested 2026-09-28)

Derived from Hlido-held evidence only (engine checklist + editorial text); quotes are verbatim from the scorecard; not vendor-supplied; re-derived daily. Verify current prices on the vendor's pricing page. Last verified 2026-09-28.

Related agents

• Claude Code Coding · 91
• Aider Coding · 90
• @aliasjeff/acli Coding · 90

Agent relevance

API CLI

Radix is a macOS app installed via a shell script that attaches to coding agents (e.g. Claude Code), giving the agent a persistent workspace in which it builds interactive artifacts wired to the repository. A free key is required, implying a backing service the agent authenticates against. The exact integration protocol (API/MCP/tool schema) is not documented on the public surface, and the app is macOS-only, so the agent-facing surface could not be inspected or exercised here.

Agent-friendly score: 6/10

Evidence

• A visual UI for agentic programming — a macOS app that gives coding agents a persistent workspace to build interactive artifacts that live in the repo and stay wired to the code — source (2026-09-28) verified
• Installs via 'curl -fsSL https://get.radix-os.com/install.sh | bash' and requires a free key obtained at signup — source (2026-09-28) verified
• Supports a broad set of agent-built artifacts (SIMD kernels, stack traces, deadlock and false-sharing analysis, test coverage, N+1 query detection, API and state-machine diagrams, feature flags, flame graphs, schemas, shaders, bisect, component maps, import cycles, login flows, rate limiters, cron views) — source (2026-09-28)