71 tools for Claude Code, Codex and Cursor

App Store optimization,
run by your agent.

Winch tracks your keywords in 175 countries on the App Store and Google Play, rewrites your listing in every language and draws your store art. You ask. It does the heavy lifting.

  • macOS 13+ · Apple Silicon and Intel
Winch's keyword board for Tidepool, a focus timer, with the agent panel open: it explains why “focus timer” is the prize and what to change in the subtitle.

Runs on the AI you already pay for

Your Claude Pro or Max plan through Claude Code, your ChatGPT plan through Codex, an OpenRouter key, or a model running locally in Ollama. No AI credits to buy from Winch and no second bill.

Countries
175
Stores
2
MCP tools
71
Download
20 MB
Telemetry
0

Keywords

Every keyword. Every store. Every day.

Neither Apple nor Google shows search volumes, and ranks are visible only by searching yourself, country by country. Winch runs those searches for you and turns them into one board: your rank with its 7-day trend, how many people a search reaches, how hard its top 10 is to beat, and what to do about it.

The keyword board sorted by rank: deep work #1, pomodoro timer #2 in four stores, focus timer #4, each with reach, difficulty and a verdict.

Competitors

Know who you’re up against.

Winch lists the apps that sit above yours on your own searches, in every country, and watches the ones you track: their ranks next to yours, every name, subtitle and version change, and new apps entering your top 10. Or type /competitors and the agent finds them for you.

Rivals: Fernwood is above Tidepool on 18 of its searches, Pomodoro Kettle on 11, each with the searches and countries where it wins.

The agent

Ask in plain words. It looks things up instead of guessing.

The agent sees everything Winch stores: every app on both stores, ranks and their history, rivals, listings, reviews, saved reports and earlier conversations. It runs new searches when it needs them, then answers with numbers you can check.

  • 65 of Winch's 71 tools, everything except the five that write to App Store Connect or Play Console. Those stay with you.
  • Stop, queue and steer it while it works. A stopped turn keeps its steps, so the next message has the context.
  • No AI credits to buy. Every answer runs on the subscription you already have and counts against that plan, not a Winch one.
  • Your provider, your model. Claude Code, Codex, OpenCode, a local Ollama, or an OpenRouter key. Nothing to set up.

Keywords

  • /ideasTen new keyword ideas, checked
  • /similarKeywords like a phrase
  • /bestThe keywords to focus on now
  • /cleanKeywords nobody searches
  • /checkCheck its ranks now

Competitors

  • /competitorsFind the apps worth watching, and track them
  • /rivalsWho outranks it, and where
  • /compareIts screenshots next to the apps that rank

Listing

  • /optimizeThree scored options for the listing
  • /localizeEvery other language, from the chosen listing
  • /descriptionWrite the App Store description

Store art

  • /captionsCaption the screenshots
  • /translateTranslate the screenshot captions
  • /iconDraw icon ideas

Release

  • /shipWhat the store still needs before review
  • /whatsnewWrite What’s New for the next version

After launch

  • /resultsWhat the last listing change did
  • /reviewsWhat recent reviews say
/
IdeasClaude Code
Every command for an app; Home adds /overview, /rivals across apps and /briefing.

MCP

Or drive Winch from the agent you live in.

The same 71 tools as an MCP server for Claude Code, Codex, Cursor and any other client. It works with the app closed, and anything it starts shows live in the window.

$claude mcp add winch -- /Applications/Winch.app/Contents/MacOS/Winch --mcp
claude
> Localize Tidepool's listing for Germany and Japan, then show me the new keyword space.
⏺ winch · get_optimization(app: "Tidepool")
⏺ winch · localize_listing(app: "Tidepool", locales: ["de-DE", "ja"])
⎿ de-DE: “Fokus Timer” tested against 4 local phrases in DE · ja: “集中 タイマー” wins in JP
⏺ winch · store_coverage(app: "Tidepool")
⎿ US reads 10 localizations · 38 free characters · 3 repeated words
Both languages are drafted. German keeps “Tidepool” and leads with “Fokus Timer” (#2 today); Japanese uses 集中 タイマー. Nothing is written to App Store Connect until you call publish_drafts with confirm.

Apps and stores7

  • list_appsTracked apps with their store, countries and kind
  • add_appTrack an App Store or Google Play app from a link or id
  • remove_appStop tracking a competitor, once you confirm
  • find_my_appsApps in a connected account that Winch doesn't track yet
  • update_appChange an app's brief and stores, or a plan's name
  • list_marketsAll 175 countries with the languages each store reads
  • search_storeSearch either store: top results, reach and difficulty

Keywords and ranks8

  • get_boardEvery keyword with rank, reach, difficulty and verdict
  • add_keywordsTrack keywords in chosen stores, optionally localized
  • remove_keywordStop tracking a keyword in one store or all
  • run_checkRun today's searches and record rank alerts
  • rank_historyThe daily rank for one keyword in one store
  • rivalsThe apps that most often outrank yours
  • head_to_headYour rank next to each tracked competitor, per keyword
  • brainstormNew keyword ideas from your AI

Listings5

  • get_listingsPublic, App Store Connect and Play Console listings and drafts
  • save_draftSave a draft for one language, cut to the store's limits
  • set_listing_textsThe description, promotional text and What's New
  • store_coverageEvery language each store reads, repeats and free characters
  • store_versionThe App Store Connect version drafts go into

Optimize11

  • optimize_listingMeasure the missed searches, then write three scored options
  • get_optimizationThe options, their scores and each language written
  • more_optionsWrite more options, unlike the ones there
  • add_optionAdd your own option, checked and scored
  • update_optionChange an option in place; it keeps its number
  • remove_optionRemove an option
  • measure_phrasesSearch up to 30 phrases so options are scored on them
  • choose_listingChoose the main listing; it becomes the draft
  • localize_listingWrite the other languages, testing local phrases in each store
  • edit_languageReplace one language's text with your own
  • pick_local_phraseUse another tested phrase in one language

Store art7

  • get_studioCaptures, captions, look, sizes and what's drawn
  • set_studioChange captions, the look, sizes, languages and order
  • write_captionsYour AI captions each capture from what it shows
  • translate_captionsTranslate the captions into every language
  • render_screenshotsDraw every language, size and capture at the store's size
  • generate_iconsDraw icon ideas with an OpenRouter image model
  • set_iconPick an icon; writes both stores' sizes

Screenshots and reviews10

  • get_screenshotsThe ranking apps' screenshots for a search, next to yours
  • compare_screenshotsA vision model's read: patterns, score, changes, captions
  • get_screenshot_reportsSaved comparisons
  • read_reviewsRead new reviews from the stores and your consoles
  • get_reviewsStored reviews, by stars, store, version or words
  • review_statsAverage, distribution, by version, store and month
  • mine_reviewsThemes, keyword ideas, angles and fixes from your AI
  • get_review_reportsSaved review reads
  • draft_review_replyDraft a reply and keep it on the review
  • save_reply_draftKeep your own reply text as the draft

Ship and launch9

  • get_readinessWhat the store still needs before review, item by item
  • update_checklistTick the steps only the store's website does
  • create_planPlan an app that isn't in a store yet
  • get_launch_planThe plan's steps, names, winnable keywords and drafts
  • check_nameIs a name free to own in search, on both stores
  • suggest_namesName ideas within 30 characters
  • choose_namePick the launch name
  • category_leadersA category's top free chart in one store
  • link_planMove a plan onto its store record

Watching and results9

  • get_activityChanges, rank moves, releases, milestones and low reviews
  • watch_nowLook at every tracked app now
  • get_resultsThe store's numbers, the change log and before-and-after
  • get_changesThe listing change log
  • sync_resultsRead newly delivered reports from Apple or Play Console
  • write_briefingWrite this week's briefing
  • get_briefingThe latest briefing
  • app_historyAn app's daily ratings, version, installs and name
  • get_conversationsEarlier conversations with the agent

Store writes5

  • publish_draftsWrite drafts to App Store Connect or Play Console
  • upload_screenshotsUpload the drawn set and graphics
  • post_review_replyPost a reply to a review
  • set_categorySet the App Store category and a game's genres
  • enable_resultsAsk Apple to start analytics reports

Each one asks before it writes, and the in-app agent never gets them.

Also from a terminal: Winch --ask "Which keywords did I lose this week?"

Listing

Write the listing three ways. Keep the one the numbers like.

Optimize measures the searches your keywords miss, writes your name, subtitle and keyword field several ways, and scores each option against today's listing on real searches. Choose one and every other language is written from it, with local phrases tested in each language's own store.

Optimize for Tidepool: three options for the main listing; the recommended “Study and deep work” adds 23 reach for one point of difficulty and covers 14 of 21 searches.

Scored, not guessed

Reach and difficulty against today's listing, with the searches each option covers marked in its fields.

Every language each store reads

Keyword space shows the words each store sees across your languages, the repeats and the room left.

Written only when you say so

Into the App Store Connect version open for editing, or one edit in Play Console. A before-and-after first, never a submission.

Store art

Your captures, captioned and drawn in every size.

Point the Studio at your app's screenshots. Your AI writes a caption for each one from what it shows and the searches your listing targets, translates them, and Winch draws every size each store takes, on your Mac, in seconds. iPhone 6.3", 6.9", iPhone Duo and iPad on the App Store; phone, tablets and the feature graphic on Google Play.

The Studio: five captures with captions such as “Focus in deep work sessions”, drawn inside an iPhone frame on a teal background.

Reviews and screenshots

Your AI reads what users say, and what they see.

Hand it your reviews, or a competitor’s: themes with counts and real quotes, keyword ideas in your reviewers’ own words, and fixes ranked by impact. Hand it a search: it lines up the first screenshots of the apps that win, scores yours out of 10 and writes caption lines to try.

Review readThemes, quotes and keyword ideas from 157 reviews
Reviews for Tidepool: 4.1 stars from 157 reviews, and an AI read with themes such as “Calm, no upsells” (41 reviews) and “Streaks lost on a new phone” (14).
Screenshot comparisonYour set next to the apps above you, scored
AI comparison for “focus timer”: Tidepool’s set scores 7 out of 10, with what to change first and what the leading apps share.

And the rest of the job

From the first name idea to the numbers after launch.

Did the change work?

Apple’s and Google’s own numbers, with every listing change marked and the 14 days before and after.

Results: downloads from App Store search rising after a listing change, with the change and its effect listed below.

Activity

Rank moves, rivals’ listing and version changes, new apps in your top 10, milestones and new low reviews.

Activity: Tidepool reached #2 for “fokus timer”, Fernwood changed its name in Germany, Tidepool passed 100,000 installs on Google Play.

Launch planner

Check a name on both stores, find keywords you can win and draft the listing, before the app exists.

Launch plan for Driftwood: four names checked (clear, taken, in use, crowded) and 8 winnable launch keywords.

Ship checklist

Everything the store needs before review, read live from App Store Connect or Play Console.

Ship: Driftwood’s checklist, with the listing and icon done and the screenshots still to draw.

Monday briefing

What moved and why, the week’s reviews, what rivals did, and three things to do. Waiting on Home.

The weekly briefing: what moved, what competitors did, reviews, and three actions for the week.

Next to the leaders

The first screenshots of the apps that win a search, side by side with yours.

Screenshots for “focus timer”: Tidepool’s live set next to Fernwood’s and Pebble Pomodoro’s.

Yours

Runs on your Mac. Answers to you.

Your data stays on your Mac

Everything Winch learns stays on your Mac. No telemetry, no tracking, no account to create.

Your keys, your AI

App Store Connect and Play Console with your own keys. AI through the tools you already pay for. Winch has no model of its own and keeps no copy of your prompts.

Nothing writes without you

Listings, store art, the category, review replies and analytics requests each wait for your confirmation. The agent is never even offered them.

Pricing

Pay once. Keep it.

$69once

A one-time license for the whole app, activated on your Macs with a key. It works offline too, and nothing you make is ever lost.

Already have a key? Manage your licenses

  • Both stores, 175 countries
  • The agent and all 71 MCP tools, on your own AI plan
  • Optimize in every language
  • Studio and icon for both stores
  • Reviews, Results, Activity, briefings

Questions

Which AI do I need?

The one you already pay for. Claude Code with your Claude Pro or Max plan, Codex with your ChatGPT plan, OpenCode, a local Ollama model, or an OpenRouter key. Winch finds them on your Mac and runs them for you. It has no model of its own and sells no AI credits.

Do I need App Store Connect or Play Console?

No. Ranks, reach, difficulty, rivals, public listings and reviews all come from the stores' public pages. Connecting your account adds your own listings, reviews with replies, the store's numbers in Results, the Ship checklist and writing back.

Will it change my listing without asking?

Never. Every store write waits for your confirmation and shows exactly what it will change. The agent never writes to a store on its own.

What do reach and difficulty mean?

Reach is how many people a search reaches, estimated from the apps that rank for it, since neither store shares search counts. Difficulty is how hard its top 10 is to beat.

Can I track competitors?

Yes. Add any app by its link or name, or let Winch suggest the ones that outrank you. You see its ranks next to yours, and every change it makes to its listing.

Does it work for Google Play?

Yes. Ranks, listings, store art, reviews and the store's own numbers all work for Google Play apps too, side by side with the App Store.

Where is my data?

On your Mac. Winch talks only to the stores and the AI you choose.

Which Macs does it run on?

Any Mac with macOS 13 or later, Apple Silicon or Intel. Windows isn’t supported yet.

Let Winch do the heavy lifting.

Download for macOS