An AI call screener that sends a daily digest of who called and what mattered
Problem
Scam and sales calls keep piling up; some people build their own bots to answer and waste the callers' time, but regular users won't self-host and lack a daily view of which calls were junk versus worth calling back, so they either miss real calls or keep getting interrupted.
Solution
A call-forwarding service with a virtual number: AI answers unknown calls, classifies spam/scam/genuine in real time, and uses different scripts; after each call it produces a transcript, summary, and callback suggestion, then sends a daily digest. Users can take over from their phone or train the allowlist anytime.
Stops multi-line text from being truncated into accidental commands when pasted into terminal TUIs
Problem
Programmers on Windows write a 10-line plan in Notepad or an editor and Ctrl+Shift+V it into a TUI like Claude CLI, only to have one line silently turned into a hard return mid-paste, sending half the instruction; even routing through Notepad reproduces it, and the break location looks random, making it painful to debug.
Solution
A lightweight clipboard middleware: before pasting, normalize the Windows clipboard (unify newlines, merge stray CRs from Word/UltraEdit) and wrap it in the bracketed-paste protocol; after pasting, verify the send buffer and warn/restore on unexpected hard returns. Ship as a standalone tray tool and optionally a VS Code extension.
Automatically recommend the right AI model, tools, and prompts for each task
Problem
Non-technical users often pick the wrong AI model, use inappropriate reasoning effort, or write vague prompts, leading to poor results. They lack guidance on model selection and prompt optimization.
Solution
A simple chat interface: users describe their task in natural language, the tool analyzes and recommends/calls the right model (e.g., GPT-4, Claude), provides optimized prompts and tool suggestions, and allows side-by-side output comparison.
Generate a verifiable 'not AI-generated' process page for handmade work
Problem
Indie developers sharing handmade games or projects are often baselessly accused of 'AI slop', leading to suppression of individual content. Creators need a way to show process evidence to counter blind AI accusations.
Solution
A lightweight tool that lets creators upload dev logs, version history screenshots, sketches, etc., and auto-generates a shareable 'making of' page with timeline, file hashes, and other verifiable info as a credibility backing.
Let job seekers' AI agents talk directly to recruiters' agents
Problem
AI-native talent and AI-native companies lack an efficient way to connect; job seekers repeatedly introduce themselves and recruiters redundantly screen, with high intermediary communication costs. Existing job platforms aren't designed for direct agent-to-agent interaction.
Solution
Provide a lightweight protocol/skill pack so job seekers' AI agents can read job listings, organize experience and preferences, auto-apply, and consolidate feedback; recruiter agents can post jobs and auto-screen matches, with agents clarifying key questions first.
Automatically strip tracking parameters from shared links on Android
Problem
Share links from many Android apps contain privacy-sensitive parameters that can even identify the sharer, but many apps don't use system share sheets, making existing de-tracking tools cumbersome. Users need an automated, frictionless way to clean links when sharing.
Solution
Use Android Accessibility Service to automatically capture share links when the user taps share, parse and strip tracking parameters, then show a secondary share sheet; support custom rules and detect/redraw QR codes in images that contain tracking links.
Set token budgets for AI coding agents and stop runaway bills
Problem
Developers using AI coding agents often see token costs explode because agents run full test suites or re-read files repeatedly. One user burned $1,200 in a single afternoon. Most only discover the spend when the bill arrives, with no real-time control during a task.
Solution
A local cost monitoring and interception tool: tracks token usage per request in real time, lets users set budgets per task, pauses the agent and alerts when overspending, and provides per-task/project cost reports to help optimize prompts and agent behavior.
Upload a vehicle photo, pick a scene, and get a short flying-car特效 clip
Problem
The Tesla Roadster demo-style flying car footage is captivating, but ordinary users who want their own car in such a video need complex editing or AI generation skills—high barrier and costly trial and error.
Solution
A template-driven AI video site: users upload a car or motorcycle photo, choose from a few preset scenes (ramp launch, jump over a truck, street takeoff, etc.), and get a 10–15s 720p clip with sound. Preview a sample before paying to generate.
Inspect outbound data from coding AI agents and mask sensitive content
Problem
Developers using coding agents don't know exactly what code, keys, or context the agent sends to model providers, nor whether sensitive data leaks. Existing tools either only log or only redact—there's no lightweight local view for quick inspection and masking.
Solution
A local proxy or CLI that intercepts outbound requests from coding agents, shows a structured view of what's sent (code snippets, file paths, env vars), and lets developers mask sensitive fields with rules or regex. Per-project retention policies.
A free Android app for small landlords to track rooms, tenants, and rent at a glance
Problem
Small landlords (PG/co-living operators) often track rooms, tenants, and rent in notebooks or spreadsheets, leading to missed collections and lost history. Existing property management software is too heavy and expensive for someone with just a few rooms.
Solution
A lightweight Android app for rental management: record tenant info, rent cycles, and payment status per room, auto-generate collection reminders, and show simple income/expense summaries. Data stays local with optional backup, no cloud dependency.
Paste a file-sharing link and let your Synology download, extract, and file it
Problem
People often share large files via GigaFile, GoFile, and similar sites. Users must download to a computer, then upload to their NAS—a wasteful round trip. Existing download tools handle multi-part archives, extraction passwords, and resume poorly, and managing many tasks is painful.
Solution
A lightweight service running on the NAS: paste a share link and the NAS downloads directly to a chosen folder. It supports queues, pause/resume, per-task folders and extraction passwords, and auto-extracts after download, including multi-part archives, waiting for missing volumes.
Turn scattered school emails, texts, and app pings into one actionable checklist
Problem
Parents are buried under a daily avalanche of school communication—emails, texts, and app notifications from teachers, admins, PTA, and after-care programs. Information overlaps, key details are buried in verbose fluff, and important dates or changes are easy to miss.
Solution
A parent-side inbox that accepts forwarded school emails/texts, automatically categorizes them (deadlines, payments, events, urgent notices), extracts key fields into a to-do list, and lets parents filter by child or school. It doesn't replace school systems—it's a noise-reduction layer for parents.
Let Codex, Claude Code and Cursor share your logged-in Chrome in parallel
Problem
Developers switch between multiple AI coding agents; official browser plugins only serve their own agent and humans end up fighting agents over one window, while agents on cloud hosts can't reach local logged-in sessions. An open-source Chrome extension plus local bridge shows the need is real and unfilled by vendors.
Solution
A local MCP bridge that hands your existing Chrome login sessions to any AI agent in isolated windows, with per-task profiles, pause/manual takeover, cleanup after each task, and one unified MCP interface across agents.
Quiz developers on AI-written code before it ships
Problem
More and more code is written by AI agents; it runs, but the author often can't explain why. A Show HN project that pops a questionnaire before pushing to prod shows this is a real anxiety, and team leads worry about juniors merging AI output straight into main.
Solution
A lightweight GitHub App/CLI that generates 3-5 targeted questions about each PR diff (why this change, edge cases, how to debug failures), scores the developer's answers, and flags low-understanding changes for reviewers.
Auto-crop, compress and convert images to each AI generator's spec before upload
Problem
Free AI image platforms each have their own upload rules: some limit file size, others constrain dimensions, aspect ratio or format, and the platform may not process it for you. Users often pick an image only to fail the upload and have to find another tool, breaking their creative flow. Existing small tools cover only a few models.
Solution
A browser-based preprocessor that automatically crops, resizes, compresses and converts images to the target model's spec (GPT Image, Nano Banana, Qwen Image, Seedream and more), with before/after comparison and file size shown live. Drag-and-drop, paste and batch processing, no signup required.
A backlink library bots can read and submit to, with free-tier labels and submission steps
Problem
Indie developers want grok bots to post backlinks on a schedule, but backlink lists are painful: awesome lists go stale with dead links, and big spreadsheets don't show which sites are free, where to submit or what's required. Even humans struggle, let alone bots. One developer manually built a 509-site directory, proving the demand and that it is currently a manual chore.
Solution
A machine-readable backlink submission directory where each site has category, price (free/paid), submission URL, required materials and steps, exposed via /llms.txt and a JSON API for bots. It tracks submission status and auto-detects dead links so bots can close the loop.
Stop AI agents from maxing out CPU and memory on small remote boxes while editing code
Problem
Developers let AI agents SSH into remote servers to fix bugs, but any non-trivial task can max out CPU and memory on a small box, making it unusable until restart. Moving most work to the local machine avoids the freeze but burns more tokens. No tool currently rate-limits, isolates and reclaims resources during remote agent sessions.
Solution
A lightweight daemon on the remote machine monitors CPU and memory of AI agent processes, automatically throttling or pausing the agent when thresholds are exceeded and cleaning up temp processes and caches. A local panel shows resource curves and session history, with per-agent quotas and auto-recovery policies.