8 Gaps Hiding in Plain Sight on HN This Week (May 29, 2026)
Eight concrete product gaps surfaced from HN comments this week: Apple Music still won't scrobble to Last.fm, there's no Deno-style secure sandbox for Python, .NET's cross-platform UI story remains broken, no browser-integrated AI slop domain blocklist exists, FL Studio users want an AI arrangement layer for their loops, Pyrefly needs inline Any/Unknown gutter markers, LLM chats lack a user-steerable assumption-check mode, and fitness app data has no open portability standard.
Source note: X/Twitter API is returning empty results as of this run. This issue draws from HN Algolia public API comments (past 7 days). Parent thread upvote counts are used as the demand signal proxy; individual comments don't carry their own scores on HN.
1. Apple Music won't scrobble to Last.fm — and nobody at Apple has fixed it
"I wish Apple Music would support scrobbling to the Last.FM API like Spotify, but to get the ball rolling we'd need someone on the Apple Music team who can nudge it up the list." — HeckFeck 1
Thread: "Last.fm is now independent" — 794 points, 48 top-level comments 2
Spotify natively supports Last.fm scrobbling. Apple Music does not, and hasn't for years. Third-party workarounds (Last.fm's native iOS app scrobbles in the background, apps like Scrobbles or Last.fm Scrobbler exist) require a separate always-running process — clunky and battery-intensive. The real ask is a first-party integration, just like Spotify's in-app toggle.
Competitive gap: Partial. Tools like "Last.fm for iOS" cover the basics, but they're unreliable after Apple system updates and require keeping a background process alive. No reliable, maintained, open-source Mac/iOS bridge exists.
Feasibility: Medium. A macOS menu-bar daemon that listens to the Apple Music Now Playing API and forwards scrobbles to Last.fm would work today and has been built before — the problem is reliability through OS updates. A better path: build a watch-face-style approach using Apple's MusicKit JS/Swift and submit to the App Store as a proper scrobble sync tool.
コンテンツカードを読み込んでいます…
2. There's no secure-sandbox runtime for Python the way Deno is for JavaScript
"I literally just discovered Deno today. I wish there was Deno for Python / WASM path was really mature. Maybe I'm missing something here, but trying to secure both a Python runtime and JS runtime for AI." — danielcasper 3
Thread: "Deno 2.8" — 418 points, 186 comments 4
Deno's value proposition is simple: run JavaScript with explicit, declarative permissions — network, filesystem, environment variables. Nothing happens you didn't allow. Python has no equivalent.
subprocess, file access, network calls — all open by default. For AI agents that execute generated Python code, this is a real security gap. Existing solutions: PyPy sandboxes (abandoned), RestrictedPython (incomplete), Pyodide (runs in browser WASM, not server-side). None covers "run arbitrary Python with Deno-style capability permissions on a server."Competitive gap: Large. Pyodide comes closest (WASM-isolated Python in the browser), but server-side permission-scoped Python execution doesn't exist in a maintained, production-ready form.
Feasibility: Medium-High. A thin permission wrapper around CPython using seccomp-bpf on Linux (like how Chromium sandboxes renderers) could get 80% of the way there. Or: lean into Pyodide and expose it as a server-side code execution API. The AI code execution boom makes this commercially viable right now.
3. .NET still has no coherent cross-platform UI story
"I wish there was a good story for .NET UI on desktop, mobile, and web. I find Maui Blazor Hybrid to be pretty close... Pretty clunky though last time I tried, and that was three years ago." — noveltyaccount 5
Thread: ".NET (OK, C#) finally gets union types" — 237 points, 279 comments 6
.NET 11 just landed discriminated union types — a long-awaited language feature. But the ecosystem gap that keeps showing up in threads like this isn't language features: it's UI. MAUI is Microsoft's latest attempt; .NET MAUI Blazor Hybrid lets you share Blazor components across desktop, mobile, and web. The problem: it's notoriously rough to set up, has inconsistent behavior across targets, and the DX is nothing like the "write once, run anywhere" promise. Flutter (Dart), Compose Multiplatform (Kotlin), and React Native fill this niche for other ecosystems. C# has no equivalent with comparable polish.
Competitive gap: Partial but painful. MAUI exists; it's just not good enough for many teams. There's a real opportunity in a "MAUI wrapper toolkit" that smooths the sharp edges — better hot reload, unified theming, and a component library that doesn't feel like a WinForms flashback.
Feasibility: Medium. Building on top of MAUI rather than replacing it keeps the scope manageable. A Tailwind-style utility-first component library for MAUI Blazor — with real docs and a showcase — could be a strong indie product for the .NET community.
4. An internet-wide "block this AI slop site forever" button
"I wish there was an internet-wide 'don't show again' button for such slop pages." — coldtea 7
Thread: "AI slop is killing online communities" — 834 points, 734 comments 8
The ask is simple: when you land on an AI-generated content farm, you want to block the entire domain from ever appearing in your search results or feed — permanently, across devices. Google's "block site" experiment from 2011 was killed. uBlacklist (browser extension) lets you manually blacklist domains from Google search results and can sync via Google Drive. But it's manual, individual, and doesn't aggregate signal across users. A shared crowdsourced blocklist for AI-generated domains — think uBlock Origin-style subscriptions for slop — doesn't exist in a polished form.
Competitive gap: Real. uBlacklist exists for manual personal blocking. There's no collaborative, automatically-updating, community-curated AI slop domain registry with browser extension integration.
Feasibility: High. Browser extension + public crowdsourced blocklist API + simple web UI for submitting/upvoting domains. The technical bar is low; the hard part is moderation (avoiding false positives on legitimate sites). Could be open-source with a hosted premium subscription for sync.
コンテンツカードを読み込んでいます…
5. FL Studio users want an AI arrangement assistant — not a new DAW
"I wish there was something like this for flstudio. I've got 25 years of loops that basically to finish them need better arrangements. Using AI to auto generate sections is what I'm missing." — tim-projects 9
Thread: "Show HN: Ableton Live MCP" 10
A Show HN project demoed an MCP server that lets Claude control Ableton Live via natural language. The comment thread surfaced an obvious corollary: FL Studio users want the same thing. But the deeper ask is specific — not "generate a whole track from scratch," but "take my existing loops and write me an arrangement structure." Verse, chorus, breakdown, outro — the arrangement skeleton that turns a loop collection into a finished track. Suno and Udio generate full tracks from text prompts; they don't ingest your stems or loops. BandLab has AI tools but they're superficial. Nobody has built "give me your loops, I'll arrange them."
Competitive gap: Large. The Ableton MCP (via Claude) partially addresses this for Ableton users. FL Studio's native scripting (Python API) is available but has no LLM-integrated arrangement layer. No tool accepts user loops as input and outputs a timestamped arrangement plan.
Feasibility: High. FL Studio exposes a Python scripting API. A thin MCP server wrapping that API + an LLM backend that understands music structure is within reach of a solo developer. The market is enormous: millions of FL Studio users, many with years of unfinished loops.
コンテンツカードを読み込んでいます…
6. Pyrefly should highlight Any / Unknown type-inference gaps in the IDE gutter
"I wish there was a config to error when an inferred type is either Any or Unknown so we can spot these locations. The report has it, but seeing it in the IDE would be great." — 5Qn8mNbc2FNCiVV 11
Thread: "Pyrefly v1.0 is here (fast type checker and language server for Python)" 12
Pyrefly just shipped v1.0 — Meta's fast Python type checker, built in Rust. Mypy and Pyright both treat
Any propagation as a normal, expected part of gradual typing. If a function returns Any, callers inherit that Any without warning. This silently poisons type safety across a codebase. The ask: a mode where the type checker flags all Any / Unknown inference points inline in the editor, not just in a report. Pyright has a strict mode (strictModes.py) and a reportUnknownVariableType diagnostic; it's close but not surfaced prominently by default.Competitive gap: Partial. Pyright's strict mode covers this if configured. But it's opt-in, rarely enabled, and buries the config. A VS Code extension that shows a subtle gutter annotation for every
Any-typed symbol — without requiring strict mode globally — doesn't exist.Feasibility: High. A VS Code language client plugin that reads Pyrefly or Pyright's LSP diagnostics and adds gutter icons/annotations for
Any/Unknown symbols is a weekend project. Publishing it could also make for a useful contribution to the Pyrefly official extension.7. LLM chat needs a proper scratchpad mode
"I wish I knew how to make Claude regressively verify its assumptions, like a kind of hook firing before a sentence is written, or after and then corrected." — MaxikCZ 1
The pattern here — also echoed by jwpapi in the same thread about agents bloating codebases to 140k lines — is that LLMs don't have a way to "think before committing." Extended thinking modes (Claude, o1) exist but they're not user-configurable. What people want: a way to force the model to explicitly state its assumptions about a problem before producing output, then let the user confirm or correct them before the model continues. Think of it as a pre-flight checklist for LLM responses.
Competitive gap: Real. Extended thinking models (Claude 3.5 Sonnet, o3) do chain-of-thought reasoning, but users can't pause mid-thought and inject corrections. No major chat UI exposes a "review assumptions before I answer" mode. This is different from system-prompt instructions — it's a structural UI feature.
Feasibility: Medium. Building a wrapper around the Claude API that uses a two-step call — first generate a bulleted list of assumptions, stream it to the user for approval, then continue with a "confirmed assumptions" context injection — is feasible as a browser extension or standalone chat client. The tricky part is making the UX feel fast enough that users don't skip the step.
8. No open data standard for fitness app portability
"I wish there was an open standard on the fitness app data for better interop." — brianjlogan 13
Fitness data is locked. Strava, Garmin Connect, Apple Health, Whoop, Oura — each runs a proprietary data silo. Apple HealthKit and Google Health Connect let apps write data to a local store, but exporting structured historical data for analysis or migration is a manual CSV download at best. The FIT file format (Garmin) and GPX are low-level protocol formats, not high-level standards. There's no "Fitness Data Portability Layer" analogous to OAuth for auth or ActivityPub for social graphs.
Competitive gap: Large. FIT SDK, OpenTracks (Android), and Health Auto Export (iOS) all exist as workarounds. But there's no open, maintained, cross-platform API spec that fitness apps can self-certify against — nothing like what Open Banking regulation achieved for financial data in the UK/EU.
Feasibility: Medium. Writing a spec is cheap; getting adoption is hard. The indie-builder angle: build a personal fitness data aggregator that polls APIs from Strava, Garmin, Apple Health, and Whoop, normalizes them into a unified schema, and exposes that as a local API. Don't try to become the standard — just solve the problem for users who want their own data. Niche, but the market is enthusiasts who will pay.
No Twitter data available this issue. Primary source: HN Algolia API (past 7 days). HN engagement counts (upvotes on parent threads) are used as demand signal proxies in place of tweet like counts. Next issue will retry X API.
参考ソース
- 1HN: Last.fm is now independent
- 2HN story: Last.fm is now independent
- 3HN: Deno 2.8
- 4HN story: Deno 2.8
- 5HN: .NET (OK, C#) finally gets union types
- 6HN story: .NET union types
- 7HN: AI slop is killing online communities
- 8HN story: AI slop is killing online communities
- 9HN: Show HN: Ableton Live MCP
- 10HN story: Ableton Live MCP
- 11HN: Pyrefly v1.0 is here
- 12HN story: Pyrefly
- 13HN: Ask HN: What are you working on? (May 2026)
このコンテンツについて、さらに観点や背景を補足しましょう。