<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:dc="http://purl.org/dc/elements/1.1/">
  <channel>
    <title>DEV Community: Tommy</title>
    <description>The latest articles on DEV Community by Tommy (@tommy_419b3e48f5ee589bbc7).</description>
    <link>https://dev.to/tommy_419b3e48f5ee589bbc7</link>
    <image>
      <url>https://media2.dev.to/dynamic/image/width=90,height=90,fit=cover,gravity=auto,format=auto/https:%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Fuser%2Fprofile_image%2F4078411%2F316f31c7-1394-4b6c-9cac-95a9565f27d4.png</url>
      <title>DEV Community: Tommy</title>
      <link>https://dev.to/tommy_419b3e48f5ee589bbc7</link>
    </image>
    <atom:link rel="self" type="application/rss+xml" href="https://dev.to/feed/tommy_419b3e48f5ee589bbc7"/>
    <language>en</language>
    <item>
      <title>Finding Deleted Tweets with Public Archives: What Works and What Doesn’t</title>
      <dc:creator>Tommy</dc:creator>
      <pubDate>Mon, 02 Nov 2026 02:23:37 +0000</pubDate>
      <link>https://dev.to/tommy_419b3e48f5ee589bbc7/finding-deleted-tweets-with-public-archives-what-works-and-what-doesnt-5jo</link>
      <guid>https://dev.to/tommy_419b3e48f5ee589bbc7/finding-deleted-tweets-with-public-archives-what-works-and-what-doesnt-5jo</guid>
      <description>&lt;p&gt;You open a bookmarked tweet and find that it is no longer available. Maybe it contained a useful explanation, a project announcement, or a reference you wanted to revisit. You still have the link, but the content is gone from the page.&lt;/p&gt;

&lt;p&gt;Can you find it again?&lt;/p&gt;

&lt;p&gt;Sometimes. A public web archive may have captured a copy while the post was accessible. Whether that copy includes readable text, images, or video is a separate question.&lt;/p&gt;

&lt;p&gt;I’m the creator of &lt;a href="https://deletedtweetfinder.app/" rel="noopener noreferrer"&gt;Deleted Tweet Finder&lt;/a&gt;, a tool for searching public tweet archives. This article explains a workflow for investigating missing posts and three design choices behind the tool: previewing results, preserving sources, and showing media availability separately.&lt;/p&gt;

&lt;p&gt;The same checks are useful when working directly with archived pages.&lt;br&gt;
An archived tweet is not necessarily a deleted tweet&lt;br&gt;
There are two different questions here:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;Did a public archive capture this post?&lt;/li&gt;
&lt;li&gt;Is the original post currently accessible?
A snapshot helps answer the first. It does not, by itself, answer the second.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;Likewise, an unavailable original page does not establish that its author deliberately deleted the post. Account availability, access restrictions, and platform behavior can complicate that conclusion.&lt;/p&gt;

&lt;p&gt;For research or reference work, keep the observations separate:&lt;br&gt;
A public snapshot contains this text. The original post was unavailable when checked.&lt;/p&gt;

&lt;p&gt;That statement is more precise than “the author deleted this.”&lt;/p&gt;

&lt;p&gt;Coverage is another limit. Public archives contain snapshots collected at particular times. They are not guaranteed backups of every account. An empty search means the available sources did not return a usable result; it does not prove that a post never existed.&lt;br&gt;
A practical workflow for finding a missing post&lt;/p&gt;

&lt;ol&gt;
&lt;li&gt;Start with the most specific information you have
An original tweet URL gives you a particular post to investigate. Keep the full URL, including its numeric post ID.&lt;/li&gt;
&lt;/ol&gt;

&lt;p&gt;If you only remember the account, start with its username. An approximate publication date can help narrow the records you examine.&lt;/p&gt;

&lt;p&gt;Preserve the original reference before searching. A link in an old document, bookmark, or message may contain information that becomes useful later.&lt;/p&gt;

&lt;ol&gt;
&lt;li&gt;Separate discovery from recovery
Finding an indexed link is a useful first step, but it is not the same as recovering its contents.&lt;/li&gt;
&lt;/ol&gt;

&lt;p&gt;A search can identify a candidate whose snapshot is unavailable, incomplete, or missing readable text. Treat the link as a lead until you inspect what was actually retrieved.&lt;/p&gt;

&lt;p&gt;This distinction also matters in interfaces. “Records found” should not silently become “posts recovered.”&lt;/p&gt;

&lt;ol&gt;
&lt;li&gt;Check identity and provenance
Before relying on a result, compare:&lt;/li&gt;
&lt;li&gt;The account and post ID with your original reference.&lt;/li&gt;
&lt;li&gt;The displayed text with the topic you expected.&lt;/li&gt;
&lt;li&gt;The publication date with your approximate timeframe.&lt;/li&gt;
&lt;li&gt;The source snapshot with the recovered record.
Keep publication time separate from snapshot time and availability-check time. They describe different events.&lt;/li&gt;
&lt;/ol&gt;

&lt;p&gt;When the source is accessible, open it. A convenient results page helps you navigate, but the underlying source gives you a way to investigate discrepancies.&lt;/p&gt;

&lt;ol&gt;
&lt;li&gt;Inspect media independently
Readable text does not guarantee that an attached image or video survived.&lt;/li&gt;
&lt;/ol&gt;

&lt;p&gt;A snapshot might preserve an image reference without the image file. A video link might remain even though the video itself cannot be retrieved.&lt;/p&gt;

&lt;p&gt;Check each attachment rather than assuming the post is complete. If the missing media carried the main point, the text alone may not answer your question.&lt;/p&gt;

&lt;ol&gt;
&lt;li&gt;Save the content together with its source
For material you want to keep, save the recovered content, original URL, archive source, and relevant dates together.&lt;/li&gt;
&lt;/ol&gt;

&lt;p&gt;Record meaningful gaps too. A note such as “text available; attached video unavailable” helps prevent an incomplete record from later being treated as a complete one.&lt;br&gt;
Three design choices behind Deleted Tweet Finder&lt;br&gt;
These distinctions shape how the project presents results.&lt;br&gt;
Preview before unlocking&lt;br&gt;
A record can be technically retrievable and still be irrelevant to the person searching.&lt;/p&gt;

&lt;p&gt;The tool provides free search and previews so users can assess what was found before deciding whether to unlock a record. Unlocking requires signing in; each newly unlocked record costs one credit, and previously purchased records do not incur another unlock charge.&lt;/p&gt;

&lt;p&gt;The useful design principle is broader than pricing: let people inspect enough evidence to make an informed decision. A result count alone cannot tell them whether their particular question has been answered.&lt;br&gt;
Keep sources attached to results&lt;br&gt;
Recovered text is easier to evaluate when its origin stays visible.&lt;/p&gt;

&lt;p&gt;Source links let readers compare the extracted content with the archived page, inspect the surrounding material when available, and retain a reference for later checks.&lt;/p&gt;

&lt;p&gt;They also make uncertainty easier to discuss. If a result looks incomplete, the reader has somewhere to investigate beyond the tool’s presentation.&lt;/p&gt;

&lt;p&gt;A source link is useful evidence, though it does not guarantee that every detail or the full conversation was preserved.&lt;br&gt;
Show text and media availability separately&lt;br&gt;
A single “recovered” label can hide too much.&lt;/p&gt;

&lt;p&gt;Text, images, and video may have different outcomes. Presenting those outcomes separately helps users understand what they can actually read, view, and export.&lt;/p&gt;

&lt;p&gt;The project’s exports contain the content that was retrieved. Exporting cannot fill gaps in the original archive.&lt;/p&gt;

&lt;p&gt;More generally, whenever an application combines several upstream resources, its status display should reflect which parts succeeded.&lt;br&gt;
An illustrative walkthrough&lt;br&gt;
The following is an operating example, not a report of a verified recovery.&lt;/p&gt;

&lt;p&gt;Suppose an old document contains a tweet link that now leads to an unavailable page.&lt;/p&gt;

&lt;ol&gt;
&lt;li&gt;Copy the original URL into the search interface.&lt;/li&gt;
&lt;li&gt;Inspect any returned records and match the post ID.&lt;/li&gt;
&lt;li&gt;Read the preview and open the archive source when accessible.&lt;/li&gt;
&lt;li&gt;Check whether referenced attachments are actually available.&lt;/li&gt;
&lt;li&gt;If the record is useful, save the recovered content and its source information.
If nothing usable appears, record that outcome without drawing a stronger conclusion. The search has not established whether the author deleted the post or whether another source holds a copy.&lt;/li&gt;
&lt;/ol&gt;

&lt;p&gt;If text appears but an image is missing, preserve that distinction in your notes.&lt;br&gt;
A useful result includes its limits&lt;br&gt;
Finding an old post is only part of the task. Understanding what was preserved makes the result useful.&lt;/p&gt;

&lt;p&gt;Public archives can provide valuable references, but their gaps need to remain visible. A careful workflow preserves identifiers, checks sources, distinguishes timestamps, and treats unavailable media as missing evidence.&lt;/p&gt;

&lt;p&gt;For developers building similar tools, those details belong in the interface from the beginning. They help users decide what they can reasonably conclude from a result.&lt;/p&gt;

&lt;p&gt;When you work with archived web content, what do you save alongside it to make it useful months later?&lt;/p&gt;

</description>
      <category>productivity</category>
      <category>sideprojects</category>
      <category>tools</category>
    </item>
    <item>
      <title>NYT Spelling Bee</title>
      <dc:creator>Tommy</dc:creator>
      <pubDate>Sat, 29 Aug 2026 08:14:10 +0000</pubDate>
      <link>https://dev.to/tommy_419b3e48f5ee589bbc7/nyt-spelling-bee-35np</link>
      <guid>https://dev.to/tommy_419b3e48f5ee589bbc7/nyt-spelling-bee-35np</guid>
      <description>&lt;p&gt;Play a free daily Spelling Bee word puzzle&lt;br&gt;
&lt;a href="https://nytspellingbee.me/" rel="noopener noreferrer"&gt;NYT Spelling Bee&lt;/a&gt;&lt;br&gt;
NYT Spelling Bee is an independent, browser-based seven-letter word game. This daily puzzle asks you to use the center letter in every word, reuse letters as often as you like, and look for a pangram that uses all seven letters.&lt;/p&gt;

</description>
      <category>game</category>
      <category>online</category>
    </item>
    <item>
      <title>Clean Voice App</title>
      <dc:creator>Tommy</dc:creator>
      <pubDate>Sat, 15 Aug 2026 03:30:30 +0000</pubDate>
      <link>https://dev.to/tommy_419b3e48f5ee589bbc7/clean-voice-app-4hcn</link>
      <guid>https://dev.to/tommy_419b3e48f5ee589bbc7/clean-voice-app-4hcn</guid>
      <description>&lt;p&gt;Clean Voice is an AI-assisted media-cleaning workspace for creators who need clearer spoken audio without turning post-production into a specialist task. Podcasts, interviews, voiceovers, online lessons, and recorded calls are often made outside controlled studios. Recordings may contain air-conditioning hum, room ambience, traffic, microphone noise, or other distractions that make speakers harder to follow. Conventional audio-editing software can address these issues, but it often requires technical knowledge, repeated manual adjustment, and time that independent creators may not have. Clean Voice makes noise reduction and voice improvement accessible.&lt;/p&gt;

&lt;p&gt;The product accepts both audio and video files, so users can work with a standalone recording or improve a video soundtrack without moving between separate tools. Its central experience is deliberately straightforward: upload a file, select an available cleaning option, let the processing engine run, preview the result, and download the cleaned media. This sequence removes the friction associated with complex editors while giving users a clear point at which to review the output. The same process can be applied consistently to podcasts, interviews, narrated videos, and other spoken-word material.&lt;/p&gt;

&lt;p&gt;At its core, Clean Voice reduces distracting background noise while aiming to keep speech clear and natural. It supports spoken-audio enhancement for recordings whose dialogue needs greater intelligibility, and cleans video soundtracks while preserving original visual content. Supported spoken-audio workflows also include filler-word cleanup, helping users reduce unnecessary vocal fillers and mouth sounds where available. Rather than exposing isolated technical settings, the product organizes these functions around a recording that is easier to hear and more ready to publish.&lt;/p&gt;

&lt;p&gt;The platform is designed around transparent task handling. Each request has a private activity record, so users can review completed work and download results later. Audio and video processing use a shared credit balance, with costs based on the relevant media workflow. This gives users a visible relationship between submitted work and the resources it consumes. Short trial use can introduce the experience before a longer workflow requires credits, while unsuccessful tasks are not intended to consume them. Task history and credit visibility make the service suitable for repeated use.&lt;/p&gt;

&lt;p&gt;Clean Voice delivers value to independent creators and small production teams. It reduces the learning curve of professional audio tools, shortens routine cleanup, and supports more consistent sound across a body of work. A podcaster can improve an interview recorded in a noisy room; a video creator can make dialogue more intelligible without rebuilding the visual edit; and a freelancer can prepare narration or client material with fewer manual steps. The project does not claim to replace a controlled recording environment or eliminate every artifact. Instead, it reduces common distractions and improves clarity when real-world recordings are the available source material.&lt;/p&gt;

&lt;p&gt;The engineering design supports this goal with a modern, extensible web architecture. The interface is built with Next.js and React for uploads, previews, account management, and activity tracking. Media processing can use a locally packaged FFmpeg workflow or the Cleanvoice service, allowing deployments to select the appropriate provider. Uploaded inputs and generated outputs are managed through object storage integrations compatible with Cloudflare R2 and Amazon S3. Processing is task-based, separating submission, execution, completion, and download while maintaining records for each user.&lt;/p&gt;

&lt;p&gt;Beyond media processing, the project includes configurable authentication, role-based access control, payment integrations, and a credit system for operating a sustainable service. Its data layer supports SQLite, MySQL, and PostgreSQL, enabling the application to fit different development and hosting environments. These capabilities turn Clean Voice from a single-purpose demo into a foundation that can be operated, extended, and adapted. By packaging complex media cleanup behind an understandable workflow, Clean Voice makes higher-quality spoken media more accessible to the people who create and share it.&lt;br&gt;
&lt;a href="https://cleanvoice.app/" rel="noopener noreferrer"&gt;Clean Voice App&lt;/a&gt;&lt;/p&gt;

</description>
    </item>
  </channel>
</rss>
