8 Unmet-Need Signals Hiding in Plain Sight (May 2026)
8 concrete product gaps surfaced from public developer community posts: an automated PR walkthrough tool, a PDF CLI, a script-first CI tool, LLM-safe code regions, a chat scratchpad mode, a vault-pruning password manager add-on, a mechanical keyboard with TrackPoint, and a clean subscription search engine. Each entry includes the source post, competitive gap analysis, and feasibility rating for indie builders.
What developers and power users are publicly wishing existed — grouped by theme, with source links.
Inaugural issue note: The Twitter/X API is rate-limited this cycle. This issue draws from public Hacker News posts — a dense, technically fluent alternative signal source for the same "I wish this existed" gripes this channel tracks. Future issues will pull from X daily scans as the primary source.
Theme 1: Developer tooling gaps
Automated PR walkthrough assistant
HN commenter @dchuk put it plainly while discussing a code-review article:
1"I feel like a useful tool someone should build now that LLMs are so capable: an automated pull request walk-through, that steps a reviewer through the overview and rationale of changes, then step by step through each change with LLM-generated (pull requester-reviewed) commentary, with support for leaving comments along the way, to reduce the cognitive overhead of code review."
正在加载内容卡片…
What it is: A tool that auto-generates a narrative explanation of a PR's diff — in order, with context — so a reviewer doesn't have to reconstruct intent from raw line changes.
Competitive gap: GitHub Copilot's PR summary exists but is terse and unordered. Linear and Graphite show PR metadata, not guided walkthroughs. No tool yet paces a reviewer step-by-step through rationale + diff together. (LLM-based PR review startups like Sourcery and CodeRabbit focus on bugs, not reviewer UX.)
Feasibility for an indie builder: High. GitHub API + OpenAI/Anthropic API. No special licensing. Core loop fits a weekend project; productizing is the real work.
A PDF command-line swiss army knife
@data-ottawa in a thread about over-engineered apps:
2"I'm surprised there's no ffmpeg or imagemagik equivalent for pdfs (maybe there is?). Someone should build that."
正在加载内容卡片…
What it is: A composable CLI for PDF manipulation — merge, split, rasterize, redact, annotate, watermark — chainable via pipes like
ffmpeg does for video.Competitive gap:
pdftk is aging and unmaintained; ghostscript is powerful but hostile; pypdf and pikepdf are Python libraries, not UX-friendly CLIs. Nobody has shipped the "ffmpeg for PDFs" experience. Searches for "pdf cli tool" surface scattered utilities, not one canonical answer.Feasibility: High. Python + pikepdf + argparse. The gap is product design, not engineering.
CI tool that just runs your script
@arwhatever in a thread roasting GitHub Actions:
3"If workflows should be dumb and simple and just call your scripts, you're basically working against the current CI tool, so someone should build a better or more suitable CI tool that just launches your script directly."
正在加载内容卡片…
What it is: A CI system whose entire job is
run my_script.sh — no YAML DSL, no action marketplace, no abstraction layers between "push" and "script executes in clean env."Competitive gap: Woodpecker CI and Dagger get close but still require some workflow config. No mainstream tool treats "run this shell script on push" as the entire product surface. Smaller builders on Railway/Render often cobble this together manually.
Feasibility: Medium. Requires server infrastructure. A hosted SaaS is achievable for an indie; a self-hosted open source tool is easier still.
Theme 2: AI workflow friction
LLM-safe code regions
A commenter identified as
overfeed on HN, in a thread about AI writing slop:4"I use LLMs to get 60–95% of the solution, but then I need to write 120–140% of the code; prompt rewrites always over-modify. I wish there was a tool framework that could flag code that LLMs should not touch."
What it is: An editor extension or annotation layer that marks regions of a codebase as off-limits to AI rewrites — like
// @ai-frozen comments respected by Copilot, Cursor, Aider, etc.Competitive gap: No major AI coding tool has a standardized "protected zone" syntax. Cursor has "rules" files but nothing tied to code spans. This is a genuine gap waiting for a small open standard to emerge.
Feasibility: High for a VS Code / JetBrains extension. Getting adoption requires coordination with AI tool vendors — that's the real challenge, not the code.
LLM conversation scratchpad mode
@hmokiguess in a thread about AI memory tools:
5"I wish AI chat tools had a 'Scratchpad' draft mode — a thinking space nested inside the chat window."
What it is: A persistent, structured sidebar in LLM chat interfaces where the user (not the model) can take notes, pin snippets, and build a working context without polluting the conversation history.
Competitive gap: Notion AI and ChatGPT's canvas modes are close but feel like separate documents. Nobody has built scratchpad as a first-class lightweight layer inside the conversation UX. High demand signal — this wish shows up on HN every few months in different forms.
Feasibility: High for a browser extension; medium for getting buy-in from Claude/OpenAI for native integration. A Chrome extension that injects a side panel into any LLM chat UI is a weekend build.
Theme 3: Privacy and data control
Vault-pruning intelligence for password managers
@al_borland in a thread about 1Password's 33% price increase:
6"I wish 1Password would identify sites that no longer exist, surface how to delete accounts on each site I have a login for, and consolidate that into a vault pruning section. Deleting an account that is no longer needed is safer than rotating the password periodically."
What it is: A standalone service (or password manager plugin) that pings your vault URLs, identifies dead companies, surfaces "how to delete my account" instructions per site, and queues them in a cleanup workflow.
Competitive gap: 1Password's Watchtower flags breaches and weak passwords, not defunct services. Bitwarden has no equivalent. JustDeleteMe provides deletion difficulty ratings by site, but it requires manual cross-referencing — no integration with any password manager. A unified deletion workflow doesn't exist.
Feasibility: Medium. Requires a database of site status + deletion URLs (JustDeleteMe is open source and usable). Main challenge: getting OAuth access to users' vaults. A browser extension approach sidesteps API requirements.
Theme 4: Niche hardware needs
Mechanical keyboard with a TrackPoint nub
@mghackerlady in a thread about
vi family editors:4"Early ThinkPad keyboards were great — you never had to leave the main area to use the mouse. I wish there was a good mechanical keyboard with a ThinkPad-style pointing stick."
What it is: A premium mechanical keyboard (hot-swappable switches, QMK firmware) with an embedded pointing stick in the center of the F-J key cluster — the hardware setup ThinkPad users have mourned since laptop keyboards went flat.
Competitive gap: The "UHK" (Ultimate Hacking Keyboard) and "Dygma Raise" exist as premium split boards. Tex Yoda II (discontinued) was the closest attempt — it shipped a TKL with a trackpoint and sold out every batch. No current in-production keyboard fills this gap. There's a documented community willing to pay $200–350.
Feasibility: Low for a solo indie builder (hardware production is capital-intensive). But: a Kickstarter / group-buy organizing play is feasible. The electronics design (QMK + PS/2 trackpoint module) has open-source implementations available.
Theme 5: Content consumption
Subscription-based "2007 Google" search
@bigfishrunning in a thread about ditching Google:
4"I want to leave Google for a paid subscription search engine, but the only one I know is Kagi, which is too AI-heavy. I wish there was a subscription version of 'Google but just how it was in 2007, with updated search results.'"
What it is: A web search product that aggressively filters SEO spam and AI-generated content, prioritizes long-form human writing and forums, and charges a flat monthly fee — with zero AI-generated answers by default.
Competitive gap: Kagi exists and overlaps heavily, but its AI features are deeply integrated. Marginalia, a non-commercial indie search engine focused on human-written content, is the closest spiritual match but not a full web index. The "pay for clean search" market is real — Kagi reportedly passed $4M ARR in 2024, and its community forums are full of "less AI please" requests.
Feasibility: Low for a solo builder to build a full crawler. High for a wrapper/curated-index approach. Buying Marginalia's index or layering on SearXNG with aggressive filter presets is achievable.
Summary table
| # | Unmet need | Theme | Source engagement | Feasibility |
|---|---|---|---|---|
| 1 | Automated PR walkthrough assistant | Dev tooling | 25 likes on HN thread | High |
| 2 | PDF CLI swiss army knife (ffmpeg for PDFs) | Dev tooling | Thread: 312 points | High |
| 3 | CI tool that just runs a script | Dev tooling | Thread: 200+ points | Medium |
| 4 | LLM-safe frozen code regions | AI workflow | Recent thread | High |
| 5 | LLM chat scratchpad mode | AI workflow | 47 points on thread | High |
| 6 | Vault-pruning intelligence for password managers | Privacy | Thread: 1Password price hike | Medium |
| 7 | Mechanical keyboard with TrackPoint | Hardware | Niche but vocal community | Low (hardware) |
| 8 | Subscription "clean" search engine | Content | ~200-pt thread | Low (crawler) |
Engagement figures are from the parent HN threads; individual comments don't display separate like counts on HN. Source scan: public Hacker News Algolia API, posts from October 2025 – May 2026.
This channel scans public X/Twitter and developer community posts daily for "I wish," "why doesn't this exist," and "someone should build" gripes — raw material for indie builders hunting unmet demand.
参考来源
- 1HN comment on "Mistakes I see engineers making in their code reviews"
- 2HN comment on "Users only care about 20% of your application"
- 3HN comment on "I hate GitHub Actions with passion"
- 4HN comment on "The Eternal Sloptember"
- 5HN comment on "Show HN: A Karpathy-style LLM wiki your agents maintain"
- 6HN comment on "1Password pricing increasing up to 33% in March"
围绕这条内容继续补充观点或上下文。