When running dozens of Claude Code sessions in the terminal, see at a glance which one needs you
Problem
Heavy agentic-development users manage 30+ Claude Code sessions in tmux but struggle to quickly identify which session is stuck, which is waiting for input, and which is done. Sometimes they only remember a name or phrase from a conversation and can't find the right session.
Solution
A tmux-native read-only dashboard that scans all running Claude Code sessions, flags ones needing attention (waiting for input, error, done), supports full-text keyword search across conversation history, and fits vim-style navigation.
Makes phone-printed Word docs and web pages sharp and font-accurate, matching desktop USB print quality
Problem
On the same wireless printer, a USB print from a computer is razor-sharp, but printing from a phone's official app, mini program, or system print service produces jagged edges, broken thin strokes, and blurry small text. Printing Word directly from a phone forces everything into one sans-serif font and wrecks the layout. Users are repeatedly frustrated by home printers from Brother, HP, Canon, and Epson.
Solution
A multi-device remote printing gateway: run a service on a PC or NAS, have the phone upload a file, and the server re-renders it with a high-DPI bitmap engine that preserves original fonts and layout, then outputs 600/1200 DPI 1-bit to the printer. Supports common wireless printer models and Word/PDF/web pages.
Merges US stock accounts scattered across brokers into one true-return and exposure view
Problem
When US stocks are spread across multiple brokers, each broker app only shows its own slice, and portfolio trackers require manual entry of every trade. Users can't easily see total holdings, how much they actually made this year, whether they beat the market, or their true single-stock exposure after overlapping ETFs.
Solution
Aggregate multi-account holdings automatically via broker APIs or CSV imports, compute time-weighted and money-weighted returns against benchmarks like SPY and QQQ, look through ETFs to calculate real single-stock exposure, and break down returns by ticker with buy/sell markers.
Auto-dismisses repetitive macOS permission dialogs like 'Allow app to find devices on local networks' and logs what it closed
Problem
The 'Allow <app> to find devices on local networks?' popup constantly interrupts typing on macOS, and there is no global setting to disable it. Similar permission prompts for location, notifications, and Bluetooth also appear mid-focus. Users almost never want to allow these but must dismiss them manually again and again.
Solution
A menu bar utility that watches for and auto-closes specified system permission popups after a configurable delay (e.g., 1/4 second). It logs each auto-dismissed app and permission type, offers a whitelist to keep prompts you actually need, and has a one-click pause.
One user vented that the phone is great for twenty minutes of attention but terrible for two-second decisions, yet much of daily life is exactly that: approve, skip, snooze. Another asked: the mouse was made for browsing, touch for scrolling, so what's the input device for approve/reject/redirect in the agent era?
Solution
A minimal always-on desktop decision panel: funnel pending approvals, message confirmations and agent requests into a card queue with one-click approve/reject/delegate, global hotkeys and optional wrist/hardware triggers, all with an audit log so nothing slips.
Stop store pages, domains and payment accounts from being locked by ex-staff
Problem
An owner found their Facebook page had been created years ago under a manager's personal account and the manager wouldn't hand it over; the shop had the bank, booking and payments, but not the page customers actually see. Smaller versions, like a domain under an ex-employee's name or an email tied to a former staffer, are common.
Solution
An account-ownership ledger for small businesses: record who registered each public-facing account (social, domain, payments, booking), the linked email, recovery method and handover status; generate an actionable recovery/handover checklist and remind when key accounts go too long without review.
Store medical reports and health logs like a password vault, ready for AI analysis
Problem
Users with chronic conditions record symptoms and hospital test data and want it stored like a Bitwarden vault: local, optionally synced to their own cloud drive, encrypted on the way up, and easy to hand to AI for analysis. Most health apps are cloud-hosted and don't expose data.
Solution
A local-first health record vault: structured entries for symptoms, visits and reports (images/PDF), all end-to-end encrypted, sync to the user's own cloud folder; an export feature that turns a chosen date range into a clean summary ready to paste into any AI.
Compress messy real-world spreadsheets into agent-readable structured summaries
Problem
When AI agents process real business spreadsheets, a medium-sized sheet can blow past a million tokens if read fully, immediately exceeding context windows. Most agents fall back to chunk sampling, which misses rows and misreads structure, a wall spreadsheet-tool builders keep hitting.
Solution
A layer between agents and sheets: parse xlsx/csv structure, formulas, merged cells and dirty data, emit a semantically faithful compact schema (columns, stats, anomaly flags), and support on-demand drill-down into a few raw rows so agents get usable signal with small token budgets.
Compare agent behavior before and after harness changes without burning money on LLM judges
Problem
Teams building vertical AI agents find offline evals can't cover real user questions, while LLM judges on all traffic are too expensive and can only sample a tiny slice, letting regressions slip into production. After a harness change nobody knows if the agent got better or worse except by gut feel.
Solution
A lightweight release health checker: snapshot every change to prompts, tools and model config, sample production/shadow traffic to compare old vs new outputs and key metrics, use deterministic rules (structural diffs, field hits, error codes) instead of an LLM judge for first-pass triage, and only surface suspicious samples for human review.
A Zotero plugin where every AI answer traces back to a page in the paper
Problem
Researchers manage papers in Zotero, but when they hand papers to a general AI the answers sound convincing yet give no clue which paper or page they came from, forcing a manual PDF check that wastes time. Zotero itself has no AI capability.
Solution
A sidebar plugin running natively inside Zotero: ask questions about selected papers and every answer fragment is forced to carry the source page and highlight, with one click jumping back to that spot in the PDF. Supports citation tracing and excerpt collection across multiple papers.
Online SSH and SFTP that keeps your keys on your own self-hosted box
Problem
People explicitly want a web-based SSH option: too much local software, fear that AI tools with full permissions on their machine will upload private keys and passwords, plus a desire for unified key management, SFTP support, a clean UI and ideally a Docker image. Local terminals are mature, but online options that don't surrender your keys barely exist.
Solution
A one-click self-hostable (Docker) web SSH/SFTP workspace where keys and passwords live in your own instance and never touch a third party: multi-tab browser sessions, basic SFTP file management, a clean UI with fine-grained permissions, and chained proxy support for when direct connections fail.
Turn enterprise security questionnaires from deal killers into reusable assets for small SaaS
Problem
Many small SaaS deals are fine on product and price, then stall when procurement sends a security questionnaire. No one internally can answer it properly, the deal drifts into follow-up and dies quietly four months later, written off as a long enterprise sales cycle.
Solution
A reusable security questionnaire answer vault: pre-fill answers and evidence links for common questions (encryption, access control, logging, backups, SOC2 status), auto-match new questionnaires, flag gaps and generate to-dos. Export and version the answer library so small teams without security staff can respond credibly.
Pay-per-project website and copy review for freelancers, no subscription
Problem
An indie dev was paying $70/month for MarkUp just to review websites and blog posts with clients, even in months with no projects to review. The need is low-frequency and spiky, yet it's packaged as an expensive always-on subscription.
Solution
A lightweight review tool for web pages and copy: clients open a shared link, click on the page to leave annotated comments without signing up. Charge per project or per review pack instead of monthly, with multi-round version comparison and exportable conclusions.
A reproducible acceptance pipeline for people running multiple AI coding agents
Problem
More and more developers let Grok Bot, Cursor Cloud Agent and similar tools edit repos and open PRs, but the fragile step is acceptance: green unit tests can still drift from the business intent, vague requirements send agents off the rails, and rework eats all the promised speedup. When juggling several sessions, people lose track of which one is stuck on approval or went sideways.
Solution
Chain requirement clarification, agent execution and reproducible acceptance into one replayable ticket: each task stores the original requirement, acceptance script, diff snapshot and result. A flat multi-session view highlights sessions stuck on approval or drifting, and a failed acceptance can be rolled back with a diff report attached.
Reusable store screenshot templates that re-render for every device and locale
Problem
Indie devs say making App Store screenshots is the most frustrating part of shipping, taking weeks to get something that looks right; every copy change, new language or new product means redoing the whole set, and required sizes differ per store.
Solution
Define a screenshot layout in the browser (background, device frame, headline, arrow callouts) with variable copy and multiple languages; fill store info once and batch-export every required size plus feature graphics, then tweak one line and regenerate all, keeping the brand consistent.
A follow-up rhythm for every quote that never says 'just checking in'
Problem
Small business owners send a quote, hear nothing back, and don't know how many times to follow up; the second message ends up being 'just checking in'. So they either nag and lose the client or give up. The same uncertainty shows up in chasing unpaid invoices: who to chase first each day is pure guesswork.
Solution
A lightweight quote and receivables ledger: record client, amount, sent date and last contact, and it generates a follow-up cadence (day 3/7/14) with a genuinely useful line for each touch (add a case study, confirm budget, give a deadline reminder). Every morning you get a 'who to follow up with and what to say' list.
Turn the engineer-needed setup for every new customer into a reusable checklist
Problem
SaaS builders report that almost every new customer needs an engineer to configure permissions, import data or tweak settings; it's fine for the first few, but onboarding soon competes with product work for engineering time. Worse, nothing is written down, so each handover starts from scratch.
Solution
Model 'one customer go-live' as a reusable checklist template: permissions, data import, integration toggles, acceptance items, each with notes and an owner. Customers fill a structured intake form, the system generates the internal runbook and a progress page, and frequent steps get exported as self-serve guides to pull engineers out of onboarding.
Catch scheduled jobs that keep slipping past their planned time
Problem
A developer found their GitHub Actions daily job scheduled at 03:17 UTC actually starting 6+ hours late every day, and drifting 20+ minutes later each day, because heavy abuse of free runners congests the schedule queue. Silent drift is almost impossible to notice yourself; you only find out when downstream data is wrong or a deploy lands late.
Solution
One job only: add a lightweight heartbeat to each external scheduled task via a curl ping or by reading public run history, compute drift between actual and expected intervals, and alert via email/Telegram/Slack past a threshold, with a drift curve and a 'time to move off this scheduler' hint.
Annotate every prospect with timezone, language and best-time-to-send
Problem
Solo founders doing overseas cold outreach complain that DMs get ignored and AI-polished English reads as obvious AI slop that recipients actively dislike. The real cause is often timing and language, not copywriting: the message lands at 3am local time, in the wrong language, and the sender is just guessing. Every bad guess burns a list.
Solution
Upload a CSV of names, companies, domains and countries; the tool infers timezone, local language, current local time and an ideal contact window, then tells you whether to send now, in which language and tone, and how long since last touch. Batch grouping and reminders only; no auto-sending, so it stays a decision layer before sending.
A tiny Git change and commit viewer built for the AI-coding era
Problem
Developers keep an IDE open next to Codex just to review local changes and history because Codex diffs are hard to use. A second IDE eats memory and spikes CPU in the background; free or open-source alternatives are either paid, ugly or bloated.
Solution
A very lightweight cross-platform Git viewer focused on uncommitted changes, commit history and file-level diffs. It lives in the tray with low memory, refreshes as AI tools edit files, and filters by folder or author.