MyMenuMind

MyMenuMind is a native macOS menu bar app for mymind. It lets you search your mind, open the 10 most recent items, and save a quick note without opening a browser.

Features

• Search mymind with the same query syntax used by access.mymind.com, including shortcuts like tag:reading, type:image, "exact phrase", cats || dogs, shoes -red, object:car, text:car, and format:pdf.
• Show the top 10 recent items returned by the mymind objects API.
• Open the best available target for an item: raw asset URL, original source URL, top-level URL, then mymind object URL.
• Save quick notes as text/markdown mymind objects.
• Store credentials in the macOS Keychain instead of source files or UserDefaults.

Requirements

• macOS 14 or newer
• Swift 6 toolchain / Xcode command line tools
• A mymind Extensions access key

Run Locally

swift run MyMenuMind

The app runs as a menu bar accessory and does not show a Dock icon. Click the brain icon in the macOS menu bar to open it.

To create a local .app bundle:

Scripts/package-app.sh

The bundle is written to .build/MyMenuMind.app.

Configure

Open Settings in the popover and set:

• Access key ID: the kid from the mymind Extensions access key.
• Access key secret: the base64-encoded secret from that access key.
• Base URL: defaults to https://api.mymind.com.
• User-Agent: defaults to MyMenuMind/0.1.0.
• API Version: defaults to 0.1.
• Object URL template: defaults to https://access.mymind.com/objects/{id}.

The access key ID and secret are saved in macOS Keychain with device-local accessibility. Other settings are saved in UserDefaults.

Use the narrowest key permissions you can:

• Search and recent items require read access.
• Quick notes require write access.

API Behavior

The implementation follows the official mymind API docs at https://access.mymind.com/api. When checking those docs from automation, use a real GET with browser-style headers; HEAD returns 405 Allow: GET.

• Every request signs a fresh HS256 JWT using the access key secret.
• JWT header: { "alg": "HS256", "kid": "<access key id>" }.
• JWT claims include uppercase method, request path without query string, iat, and exp.
• The signed JWT is sent as Authorization: Bearer <jwt>.
• User-Agent is sent on every request.
• API-Version: 0.1 is sent on every request so the beta API is pinned.
• Search uses GET /search?q=<query>&limit=<limit>, then resolves matched IDs with GET /objects?id=<id>....
• Recent items use GET /objects?limit=10&contentAs=text/markdown, assuming the API returns the most recently bumped objects first. The app keeps API order for equal or missing timestamps.
• Quick notes use POST /objects.

Quick notes are sent as JSON:

{
  "content": {
    "type": "text/markdown",
    "body": "Your note"
  }
}

The item parser accepts common response envelopes:

• top-level arrays
• { "items": [...] }
• { "data": [...] }
• { "data": { "results": [...] } }
• { "results": [...] }
• { "cards": [...] }

Verify

swift test
Scripts/check-release-ready.sh

The release check runs tests, builds the app bundle, ensures local-only artifacts are not in the publishable file set, and scans publishable files for common secret patterns.

Public Publishing Checklist

Before pushing this repository publicly:

1. Run Scripts/check-release-ready.sh.
2. Confirm git status --short --ignored shows .env, .claude/, .build/, and mymind-api-docs/ as ignored.
3. Confirm only source, tests, docs, scripts, resources, and GitHub workflow files are staged.
4. Never paste a real mymind access key into issues, pull requests, commit messages, logs, or docs.

Publish Hygiene

This repository intentionally ignores local and generated files:

• .env and .env.*
• .claude/
• .build/
• .swiftpm/
• mymind-api-docs/

Do not commit API keys or generated app bundles. If a mymind key is exposed anywhere public, rotate it immediately.

Archive API Docs

The scraper and audit tools are kept in Scripts/; the generated archive is ignored by Git.

python3 Scripts/scrape-mymind-api-docs.py --output mymind-api-docs
python3 Scripts/audit-mymind-api-docs-with-agent-browser.py mymind-api-docs

The scraper uses real GET requests with browser-style headers, crawls every discovered https://access.mymind.com/api... page, downloads docs-owned static assets from static.accelerator.net, and writes manifest.json, raw HTML/assets, extracted page text, and all-content.txt.

Coverage behavior:

• Starts at https://access.mymind.com/api.
• Probes robots.txt, sitemap.xml, /api/sitemap.xml, and /api/llms.txt for additional route discovery.
• Parses HTML links, HTML comments, CSS url(...), and string URLs inside downloaded HTML/CSS/JS.
• Recurses through every discovered /api... page until the queue is empty.
• Fails the run if max_urls is reached before queues drain.
• Records crawl_completed, unfetched_discovered_pages, hashes, statuses, and failures in manifest.json.

The Agent Browser audit writes rendered/, snapshots/, snapshots-full/, agent-browser-audit.json, and all-rendered-content.txt. Prefer all-rendered-content.txt or snapshots-full/ when you need what Agent Browser sees, because some code examples and the LLM instructions block are injected by JavaScript after the raw HTML loads.