FAQ & Troubleshooting

Chrome:

1. Download or unzip the extension folder to a permanent location on your computer (e.g., Desktop or Downloads). Do not delete this folder after loading — Chrome reads directly from it.
2. Open Chrome and type chrome://extensions/ in the address bar. Press Enter.
3. In the top-right corner, flip the Developer mode toggle to ON.
4. Click the Load unpacked button that appears.
5. Navigate to the screenshot-extension-6 folder and select it, then click "Select Folder."
6. SwiftShot now appears in your extensions list with its camera icon.

Edge:

Same steps, but go to edge://extensions/ instead. The Developer mode toggle is in the bottom-left corner.

1. Look for the camera icon in your browser toolbar (top-right area, next to the address bar).
2. If you don't see it, click the puzzle piece icon (Extensions menu) and pin SwiftShot by clicking the pin icon next to it.
3. Click the camera icon to open the popup with three capture options:
   • Visible Area — captures exactly what's on screen
   • Select Region — click and drag to capture a specific area
   • Full Page — scrolls the entire page and stitches it into one tall image

The editor opens automatically after every capture. When you click any capture button:

1. The extension captures the screenshot.
2. A new tab opens with the annotation editor loaded.
3. Your screenshot appears on the canvas, ready to annotate.

There is no separate "open editor" step — it's always triggered by a capture.

1. Go to chrome://extensions/ (or edge://extensions/).
2. Find SwiftShot in the list.
3. Click the refresh icon (circular arrow) on the extension's card.
4. Close and reopen any editor tabs that were already open.

Very long pages (40+ screen heights) hit browser canvas limits. The extension caps at 40 scroll captures and 32,768px canvas height.

For extremely long pages, try:

• Capturing in sections using Select Region mode
• Exporting as a multi-page PDF which handles pagination automatically

The extension detects and handles sticky headers, footers, and sidebars automatically. If you see duplication:

• Refresh the page first, then recapture
• Some pages with unusual CSS layouts may not be detected correctly

Make sure the page is fully loaded before capturing. Infinite scroll pages (social media feeds) may take longer. Wait for the editor tab to open — it can take 10-30 seconds on very long pages.

Free users have a daily capture limit of 5 screenshots/day. The limit resets at midnight. Upgrade to Pro for unlimited captures.

This happens when the OCR engine times out. Common causes:

• Very long screenshots — the extension downscales images over 8 megapixels automatically, but extremely large captures may still time out. Try capturing a shorter section.
• Image-heavy pages with little text — OCR works best on text-heavy content.
• Browser memory pressure — close other tabs to free up memory, then recapture.

The timeout scales dynamically (30s base + 15s per megapixel, up to 3 minutes).

Tesseract.js runs entirely in your browser (no cloud). Accuracy depends on:

• Font size — very small text (below 10px) is harder to recognize
• Image quality — blurry or low-contrast text produces worse results
• Language — currently only English is supported

After capturing a screenshot from a web page:

1. Look for the purple Quick Summary button in the top toolbar (next to Speed Read).
2. Click it to open a full-screen overlay with a 3-sentence extractive summary, top keywords, word count, and reading time estimates.
3. Use Copy Summary or Copy Full Text to grab text. Click the X or press Escape to close.

The button only appears when text is available. This requires either:

• Article text extraction — automatically extracts text from the page's HTML when you capture (works on most news sites, Wikipedia, blogs, and docs).
• OCR completion — if article extraction didn't find enough text, the button appears after OCR finishes.

If you captured a local image file or a page with only images, neither source will produce text and the button won't appear.

The TextRank algorithm ranks sentences by their similarity to other sentences in the document — not by position. If the summary feels intro-heavy, it's because the introduction genuinely contains the most interconnected concepts.

The algorithm also:

• Filters out short fragments and headings
• Penalizes reference/citation lines
• Strips Wikipedia noise ([1], [edit], [citation needed])

Yes — if OCR successfully detects text in your screenshot, the summary button will appear regardless of the source. DOM-based article extraction only works on live web pages, but OCR works on any image.

1. Capture a web page (any mode).
2. In the editor, click the Speed Read button in the top toolbar.
3. A full-screen overlay shows one word at a time with the Optimal Recognition Point highlighted in orange.
4. Use speed presets (Slow / Normal / Fast / Sprint) or the slider to adjust WPM.
5. Press Space to pause/resume. Press Escape to exit.

Same as Quick Summary — it requires text from either article extraction or OCR. It appears alongside the Quick Summary button once text is loaded.

In the editor, click the microphone button in the left sidebar toolbar. A listening overlay appears. Speak commands like:

• "Draw a red arrow" / "Add blue rectangle"
• "Highlight" / "Blur" / "Redact"
• "Undo" / "Redo" / "Delete"
• "Save" / "Export PDF"

Troubleshooting steps:

• Allow microphone access — your browser will prompt for permission the first time. Click "Allow."
• Check browser support — voice requires the Web Speech API (Chrome and Edge only, not Firefox or Safari).
• HTTPS required — voice recognition only works on HTTPS pages or extension pages.
• Zoom conflict — if zoomed in (zoom > 100%), Space is reserved for panning. Click the mic button directly instead.

• Speak clearly at a normal pace
• Use exact phrases: "red arrow" not "put a reddish arrow thing"
• Reduce background noise
• Commands are capped at 200 characters for security

If you're zoomed in (Ctrl+scroll), coordinates are adjusted automatically. If they still seem off, reset zoom to 100% by pressing 0 (fit width) or using the zoom controls.

Look for a thin strip on the right edge of the editor with a small chevron arrow. Click it to expand the annotations panel back to full width.

Three ways:

• Press Ctrl+Z (or Cmd+Z on Mac)
• Say "Undo" if voice is active
• Keep pressing for multiple undos

Redo with Ctrl+Shift+Z.

Select the object first (click on it, or press V to switch to Select tool), then press Delete or say "Delete" with voice control.

The vault is an encrypted archive system built into the editor. It stores screenshots with annotations, OCR text, and summaries in AES-256-GCM encrypted storage on your device. Everything stays local — nothing is uploaded.

In the editor toolbar, click the vault/lock icon. You'll be prompted to set a master password (first time) or enter your existing password, then click Save to Vault.

If you set up Windows Hello biometric unlock before forgetting your password, you can still unlock with fingerprint or face. If both are lost, the data is unrecoverable.

Vault archives auto-expire after 30 days by default. When an archive is within 3 days of expiring, you'll see a warning banner with options:

• Keep Forever — removes the expiration
• Export — downloads an unencrypted backup
• Let Expire — archive auto-deletes after the expiration date

You can change any archive's expiration (7d / 30d / 90d / 1yr / Forever) from the archive viewer.

Vault passwords require 12+ characters with at least one uppercase, one lowercase, one digit, and one special character. This is intentional — the stored salt and test record would allow offline brute-force attacks against weak passwords.

Cloud Vault is an opt-in evidence preservation system for law firms and professionals who need court-defensible screenshot evidence. It stores screenshots in AWS S3 with Object Lock (WORM — Write Once Read Many), meaning evidence cannot be tampered with during the retention period.

Key capabilities:

• WORM storage — screenshots are locked from modification or deletion once uploaded
• SHA-256 hash verification — every screenshot gets a cryptographic hash at capture time
• Chain of custody log — records who captured what, when, and where
• Configurable retention — default 3 years, adjustable per case
• Governance mode — evidence can be released after a case closes, but cannot be altered during retention

Evidence is uploaded directly to AWS S3 via presigned URLs. This means:

• The screenshot travels directly from your browser to S3 — it never passes through SwiftShot servers unencrypted
• Images are encrypted at rest using S3 Server-Side Encryption (SSE)
• A SHA-256 hash is computed locally at capture time and recorded in the chain of custody log before the upload begins

1. You capture a screenshot normally.
2. The extension computes a SHA-256 hash of the image locally.
3. A presigned upload URL is generated (time-limited, single-use).
4. The screenshot is uploaded directly to S3 with Object Lock enabled.
5. The chain of custody log records the capture metadata: user, timestamp, source URL, and hash.

WORM (Write Once Read Many) means that once a file is written to storage, it cannot be modified, overwritten, or deleted until the retention period expires. AWS S3 Object Lock enforces this at the infrastructure level — not even account administrators can alter locked objects.

SwiftShot uses Governance mode, which means:

• During the retention period, evidence cannot be tampered with by anyone
• After a case closes, authorized administrators can release the lock to manage storage
• The default retention period is 3 years, configurable per case

This is critical for legal proceedings because opposing counsel can challenge evidence authenticity. WORM storage provides an infrastructure-level guarantee that the screenshot has not been altered since capture.

Every screenshot gets a SHA-256 cryptographic hash computed at the moment of capture, before any upload occurs. This hash acts as a digital fingerprint.

To verify a screenshot has not been altered:

1. Take the screenshot file and compute its SHA-256 hash (any standard tool works).
2. Compare the computed hash against the value recorded in the chain of custody log.
3. If they match, the screenshot is verified as unaltered since capture.

Anyone — opposing counsel, judges, forensic experts — can independently verify a screenshot's integrity by comparing its SHA-256 hash against the custody log. No special software is required.

The chain of custody log is an append-only record that tracks every action taken on a piece of evidence. For each screenshot, it records:

• Who — the authenticated user who performed the capture
• What — the SHA-256 hash identifying the exact image
• When — the precise timestamp of capture and upload
• Where — the source URL of the captured page

This log is essential for legal proceedings where you need to demonstrate an unbroken chain of custody from capture to courtroom presentation.

All tiers include the full local extension with capture, annotation, OCR, and encryption. Cloud evidence features are additive.

No. The extension continues to run 100% locally. Cloud Vault is an opt-in add-on:

• All captures, annotations, OCR, and local vault encryption work identically whether Cloud Vault is enabled or not
• If you enable Cloud Vault, screenshots are uploaded to S3 in addition to being saved locally
• If your internet connection drops, captures still work locally — cloud uploads queue and retry automatically
• You can disable Cloud Vault at any time without affecting your local workflow

• PNG — click the download button or press Ctrl+S
• PDF — click "Export PDF" for single or multi-page output
• Encrypted file (.enc) — for password-protected sharing
• Vault — for local encrypted archival
• Email — click the share button

The extension generates standards-compliant PDFs with byte-accurate xref tables. If a viewer shows errors, try opening in Chrome's built-in PDF viewer or Adobe Acrobat. Some lightweight readers are stricter about formatting.

In the left toolbar, look for stamp tools: Exhibit label (A, B, C...), Bates number, Confidential, Privileged, and Signature line. Click the tool, then click on the canvas where you want it placed. These are Pro features.

• Very large screenshots — the editor handles images up to 32,768px tall, but this uses significant memory.
• Many annotations — hundreds of objects slow rendering.
• OCR running — CPU-intensive but runs in a Web Worker. May cause brief slowdowns on older machines.

Try closing other tabs to free memory.

1. Go to chrome://extensions/
2. Click the refresh icon on SwiftShot
3. If that doesn't work, toggle it off and back on
4. As a last resort, remove and re-add it ("Load unpacked" again)

This is a fallback warning, not an error. The extension tries a 32,768px canvas first, and if the browser doesn't support it, falls back to 16,384px. The capture still works, but very long pages may be truncated.

Common non-critical messages:

• [SSP] Canvas height ... clamping — fallback to smaller canvas (see above)
• service worker registration failed — usually clears on reload
• Could not establish connection — normal when navigating away during capture

If the extension is working correctly, these can be safely ignored.

No. Everything runs 100% locally:

• OCR — Tesseract.js runs in your browser via WebAssembly
• Summarization — TextRank algorithm runs in JavaScript locally
• Encryption — Web Crypto API (your browser's built-in crypto)
• Vault — IndexedDB on your device only

No analytics, no tracking, no telemetry.

The vault uses:

• AES-256-GCM encryption (industry standard)
• PBKDF2 key derivation — 100,000 iterations, SHA-256
• 12+ character password policy with complexity requirements
• Rate limiting — 5 failures = 30s lockout, 10 = 5min
• Auto-disposal — 30-day default expiration

All encryption runs via the Web Crypto API — the same crypto engine used by banks and secure websites. Your password never leaves your device.