Tools

Use a readable short URL when people may hear, type, copy, or share the address. Use a QR code for print or signage when the intended action is a phone scan. The same destination can be offered both ways for accessibility and convenience.

Before distributing either, confirm the destination, test redirects, and check whether the short-link service is likely to remain available. For a QR code, scan the final printed size on more than one device. Neither format proves the destination is trustworthy; users should be able to inspect the address.

Do not put private tokens in either link.

Keep the whole code and its quiet border in the screenshot, then use an image-capable QR scanner to decode it. A tiny, blurred, or heavily compressed crop may not scan; try a clearer source rather than sharpening it aggressively.

Read the decoded text first. If it is a URL, inspect the full hostname and be wary of shortened links or lookalike domains. If the code asks for a login or payment, navigate to the organization's site independently before proceeding.

A QR code hides text from casual view; it does not make that text safe.

Work from a copy and ask what the service needs to complete the task. Remove unrelated pages, embedded metadata, credentials, personal identifiers, and confidential material where possible. Check the tool's data handling and whether the file is processed locally or sent elsewhere.

Use a small synthetic sample first if you are unsure how conversion will behave. After processing, open the exported result and inspect for missing pages, altered formatting, or residual private details. Keep the original until the output is verified.

A browser tool can be convenient; convenience alone does not establish privacy.

Create a small folder for utilities you return to, with names that describe the job: “Format JSON,” “Compare text,” or “Resize image.” Keep the links you actually use and remove duplicates.

A short list is faster than a sprawling collection. If a tool handles sensitive data, note whether you trust its processing model before making it your default. Review the folder occasionally when your workflow changes; the best bookmark is the one that saves a repeated search.

Use browser search for rare tasks instead of storing every interesting utility.

A browser utility fits a one-off conversion, calculation, or small edit when you only need the output. A full app becomes worthwhile when the work needs project history, collaboration, layers, batch processing, or repeated revisions.

Consider file size, privacy, accessibility, export quality, and how often you will return to the task. Try a small safe sample in the quick tool, then inspect the exported result. If you repeatedly reconstruct the same settings or need to preserve intermediate work, a more capable app or workflow may save time.

Choose based on the finished task, not the number of features.

Use a text diff to review paragraphs, files, and blocks where added or removed lines matter. Use a character diff for short strings, punctuation, IDs, or an invisible-space bug inside one line.

Before interpreting a noisy result, check line endings, tabs, trailing spaces, and Unicode normalization. A visual match does not always mean the underlying bytes match. Conversely, line wrapping can make a text diff look dramatic when the wording is unchanged.

Compare the smallest meaningful unit first.

An XML formatter can add indentation and line breaks so nested elements are easier to inspect. Work on a copy and compare the parsed content before and after formatting if whitespace inside text nodes matters.

Formatting does not repair missing closing tags, invalid entities, or a wrong schema. Use an XML parser to check well-formedness, then validate against the consuming application's rules. Inspect namespaces, attributes, and mixed content; pretty-printing can change meaningful whitespace in some documents.

Readable XML is not automatically correct XML.

A Unix timestamp commonly counts seconds since 1970-01-01 00:00:00 UTC; some systems store milliseconds instead. A 10-digit modern value is often seconds and a 13-digit value often milliseconds, but check the source format rather than relying only on length.

Convert it with a timestamp converter, then display it in the viewer's time zone when that is the question. The instant stays the same while the local date and hour may change. If the result lands decades away from expectation, verify seconds versus milliseconds first.

Keep UTC and local representations labeled.

Choose the clearest image: upright text, adequate resolution, strong contrast, and little glare. Run image to text, then compare the extracted text with the source instead of treating a plausible paragraph as exact.

Check names, amounts, dates, punctuation, and similar characters such as O and 0. Tables and multi-column layouts can lose reading order. For important documents, preserve the source image and verify every value used in a decision or calculation.

OCR makes text editable; it does not certify that the transcription is accurate.

Choose the destination's aspect ratio first: square avatar, portrait post, or wide banner. Crop from the best available original and keep faces, text, or product details away from edges where masks and overlays may cut them off.

Preview the exported image at the actual size and shape, including a circular avatar mask if relevant. Resize from the cropped original rather than repeatedly saving smaller copies. Keep enough surrounding context that the subject remains recognizable without becoming tiny.

The crop should serve the message, not merely eliminate empty space.

Blurry output often comes from enlarging a small source or repeatedly saving a lossy file. Begin with the highest-quality original, choose the final aspect ratio, and calculate pixel dimensions from the actual display size. A 400 CSS-pixel slot can need roughly 800 source pixels on a 2× display.

Downscale once with a good resampler, inspect fine text and edges at the intended viewing size, and export a suitable format. Photographs often work well as JPEG, WebP, or AVIF; sharp diagrams may need PNG or lossless WebP. Keep an untouched master and generate separate derivatives for different layouts. Do not expect an upscaler to restore detail that the original never captured.

Use a general unit converter for length, mass, area, and volume. Temperature needs its own formulas; data size needs a clear distinction between decimal units such as MB and binary units such as MiB. Cooking measures may depend on the ingredient, especially when converting volume to mass.

Write the original value with its unit, select the target, and inspect the result's rounding. For an important measurement, convert back or compare a known reference. Do not assume two similar unit labels mean the same thing.

Keep units attached to the numbers throughout.

Percentage calculations answer different questions. “What is 25% of 120?” gives 30. “42 out of 50 is what percent?” gives 84%. “80 increased to 96” is a 20% increase, using 80 as the base.

Choose the matching mode in a percentage calculator and check which number is the baseline. For a discount, calculate the amount removed and the final price separately. Round only at the end and state the currency or unit where needed.

The arithmetic is simple once the base is unambiguous.

A pleasing palette is not enough if small labels or controls are hard to read. Test the actual foreground and background pair with a contrast checker, including transparency, hover states, focus indicators, and dark mode.

Inspect the design at phone size and normal brightness. Use text, shape, or icons alongside color for errors and status so meaning survives limited color perception. If a brand color fails as body text, keep it for accents and choose a more readable text color.

The computed colors in the page matter more than swatches viewed alone.

Use a Markdown preview when you want to see how headings, lists, tables, links, and code blocks render as you write. Keep the Markdown source editable until the text is final.

Use Markdown to HTML when the destination specifically needs HTML. Inspect the converted output in that destination, because Markdown dialects and HTML sanitization can differ. Verify links and complex blocks rather than assuming the preview and published page will match exactly.

The choice depends on whether you are checking appearance or producing another format.

A case converter can quickly turn all-caps notes into readable copy or standardize a list of headings. Choose the intended style first, then apply it to a sample before changing the whole document.

Review acronyms, brand names, names, identifiers, and text in languages with special casing rules. Automatic title case may capitalize small words oddly; lowercasing an identifier can change its meaning. Keep the original for comparison when exact spelling matters.

The tool saves typing, but it cannot decide which letters are semantically important.

Paste a draft into a word counter when you need an approximate reading length or an essay target. Use a character counter when a field has a strict limit. Check whether the receiving platform counts spaces, line breaks, emoji, or bytes differently.

Reading-time estimates are rough because text complexity and audience vary. Use the count to guide editing, then read the actual text for clarity. For a hard form limit, paste the final version into the destination and verify it fits there.

Counting is a measurement, not an edit.

The first digit gives a broad category: 2xx success, 3xx redirection, 4xx a request the server cannot fulfil as sent, and 5xx a server-side failure. That does not mean every 4xx is the user's fault or that every 5xx is safe to retry blindly.

Read the method, URL, response body, relevant headers, and redirect chain. 401 commonly means authentication is needed; 403 means access is refused; 409 signals a conflict; 429 calls for rate-limit handling. A 5xx after a non-idempotent request may occur after the action succeeded, so check state or use an idempotency key before retrying. The HTTP status reference gives the individual meanings.

Start with a trusted breach-notification service or a notice from the affected company; use the service’s official site rather than a link from an alarming message. Never paste a live password into an unfamiliar “breach checker.” A missing search result is not proof that a password was never exposed.

If an account password may be compromised, change it through the service’s official login page. Replace the same or similar password anywhere else you used it, starting with email and financial accounts. Use unique passwords stored in a password manager and enable the strongest available multifactor authentication. Review active sessions and recovery settings if account takeover is suspected. The FTC recommends prompt changes after a password-related breach notice.

A random-looking identifier and a secret token solve different problems. For a record ID, collision resistance may be enough; for a reset link or API credential, the value must also be unpredictable to an attacker. Use a cryptographically secure generator for secrets, not timestamps, Math.random(), or a phrase you invented.

Choose enough length for the purpose, make each secret single-use when possible, and set an appropriate expiry. Keep tokens out of analytics, logs, and URL sharing previews. Store a hash of a verification token when the application only needs to check it later. The password generator is useful for human account passwords; a UUID is not a replacement for a carefully designed access token.

A CSS beautifier can restore indentation and line breaks to a minified stylesheet. Paste a copy into the code beautifier, inspect the output, and keep the original until you have confirmed the rules still parse. Do not assume formatting can recover removed comments, source variables, or the structure of Sass or another source language.

If the file came from a build, find the original source or source map before editing; a direct change to generated CSS will disappear at the next build. Preserve declaration order and duplicate properties, which may be intentional fallbacks. After a change, compare rendered pages at relevant breakpoints and run the normal build, then let the pipeline minify the result again.

JPEG is often efficient for photographs but can blur sharp text or edges. PNG preserves exact pixels and transparency, making it useful for screenshots and graphics. WebP supports both lossy and lossless modes; compare its output rather than assuming one format always wins.

First resize from an untouched master to the dimensions the page will display. Export a few candidates, then inspect fine text, gradients, and transparency at real viewing size. Watch metadata and color profiles when those matter. Repeatedly saving a JPEG adds artifacts, so generate each derivative from the master. File size is important, but the smallest file is not useful if it destroys the detail the image needs to communicate.

A common five-field cron expression describes minute, hour, day of month, month, and day of week. For example, 30 9 * * 1-5 usually means 09:30 on weekdays. The exact scheduler matters: some add a seconds field, and interpretations of restricted day-of-month plus day-of-week can differ.

Use the cron parser to check the expression, then inspect its next several run times in the scheduler's configured timezone. Test across daylight-saving changes if the job is local-time sensitive. Also decide what happens when a run is missed or overlaps the next one. A correct string is only one part of a reliable scheduled task.

A diff highlights added and removed lines, but its quality depends on what you compare. Put the known old version beside the proposed new version in the diff checker. Review each changed hunk with surrounding context, then scan the final document for inconsistencies a line-level view can miss.

If the output is noisy, check line endings, indentation, and generated timestamps. Normalize only differences that truly do not matter; whitespace can be meaningful in code or data. For JSON, a structural diff may be better than a text diff, especially when object keys moved. Keep confidential text in a trusted local tool instead of pasting it into an unknown service.

A UUID is a 128-bit identifier that applications can generate without asking a central counter for the next number. Version 4 draws from random or pseudorandom data; other versions use different designs.

Use UUIDs for record IDs, tracing, and keys when decentralised generation is useful. Validate the format and version, and keep a uniqueness constraint in a database where a collision would be harmful. RFC 9562 explains why absolute global uniqueness cannot be guaranteed.

A UUID is not automatically a secret access token. Authorization must still check who is allowed to read or change the record. Choose a purpose-built secret when unguessability is the security requirement.