Text & everyday tools · Text Toolkit
Cleaning a list of client emails online? Check where the text goes first
· Why it matters
text-tools privacy data-handling
Personal data pasted into a web tool may be sent to and logged by a server; this post explains what to check before pasting a client or staff list, and what 'runs in your browser' should mean.
The staff list you pasted into a web page — names, emails and phone numbers are personal data even when the job is trivial
A merged staff list can look like an ordinary housekeeping task: remove padded spaces, discard duplicate rows and put the addresses in order. Yet the pasted block may identify employees, contractors or clients through names, email addresses and phone numbers. The simplicity of the edit does not reduce the sensitivity of the underlying records.
Before copying the list into any web page, separate the operation from the data. A harmless sample can establish whether a tool performs the required cleanup, while the real list should be used only after its processing path is understood. Minimising the pasted fields also limits exposure when names or numbers are irrelevant to deduplication.
What a server-side tool can do with your text — logging, caching, error reports and third-party analytics, none of which need bad intent
A tool backed by a server could receive submitted text in a request and may pass it through infrastructure that logs requests, records errors or caches responses. Those are possible architectures, not accusations about every site. Good intent alone does not answer the practical question of which systems can receive or retain the list.
Third-party scripts can create additional network paths even when the main transformation appears simple. The relevant evidence is the actual request traffic and current implementation, not a generic warning or reassuring badge. Determine whether text becomes a request parameter or body before deciding that a page is suitable for personal data.
What a server-side tool could do with submitted text, without assuming any particular service does it
Ask four concrete questions: is the input uploaded, is it stored between visits, does the page load external scripts, and can the stated behavior be checked? A clear privacy statement is useful because it defines a claim, but it should identify the scope of that claim. “Browser based” should mean the transformation executes on the device.
ToolAcre’s manifest states that local processing is enabled, no account is required, analytics are disabled and no network is required after load. Its text-tool configuration says pasted content is not autosaved and disappears on reload or Clear data. Those repository facts are stronger than an unexplained lock icon, while runtime observation checks the deployed page.
Questions to ask and evidence to inspect, rather than treating a privacy label as proof
Open developer tools, select Network and clear the recorded requests after the page finishes loading. Paste a deliberately fake list, then trim each line, remove duplicate lines and sort the result. The source implements those actions as string operations: lines are split locally, trimmed with JavaScript methods, filtered through a Set and sorted with localeCompare.
Watch for new requests while each action runs and inspect their URLs, headers and bodies if any appear. Silence supports the narrow observation that the tested actions did not transmit the sample in that session. It does not prove what every browser extension or future deployment will do, so record the page, browser and time when evidence matters.
Shared and public computers — why 'nothing saved between visits' and a Clear data control matter on a hot desk
A shared computer introduces a different risk: the next person may see content left in the tab, clipboard, downloaded file or browser history. ToolAcre’s text configuration says nothing is restored after reload and that Clear data empties the text and undo history. Use that control before leaving, then close the tab rather than relying on inactivity.
Local processing does not make the screen private. Someone nearby can read it, an extension may have broad page access, and a downloaded result persists wherever the browser saves it. On a hot desk, prefer an approved profile or device, paste only necessary columns and remove the exported file according to the organisation’s handling rules.
Worked example — cleaning a merged staff list locally, exporting the result, and confirming nothing left the machine
For a controlled example, create fake rows such as ` alex@example.invalid `, `sam@example.invalid` and a repeated Alex entry. Paste them into Text Toolkit, trim line whitespace, remove duplicate lines and sort ascending. Because duplicate comparison can be case-sensitive, decide whether differently capitalised addresses should remain distinct before running the operation.
Download or copy the cleaned result, confirm it contains the expected two rows, and inspect Network for the sample addresses. The manifest declares plain-text output and the tool configurations describe a downloadable `text.txt`. Clear the editor and undo history afterward, then verify the destination file rather than assuming a successful download placed it in an approved location.
Worked example: clean and export a staff list locally, then check the observed network activity
This check does not decide whether staff or client data may be processed on a particular device. Internal policy, contractual duties, access controls and retention requirements remain outside a text transformation’s source code. Approval for one local utility also does not automatically cover another page with different scripts, ownership or storage behavior.
The exported list becomes a new copy that needs its own protection. Review its folder permissions, backup behavior, sync clients, sharing settings and deletion schedule under the rules that apply to the original records. A silent Network panel during cleanup says nothing about what a cloud-synchronised Downloads folder may do after the file is created.
The takeaway — the Text Toolkit processes the list as functions in the page with no server behind it, so the question never arises
ToolAcre’s implementation performs trimming, deduplication and sorting with functions in the page, and its manifest says text never leaves the page, analytics are disabled and no network is required after load. That supports a specific claim about the shipped design: the cleanup operation itself has no conversion server receiving the list.
Treat that design evidence as one part of a responsible workflow. Test with fake data, inspect runtime requests, minimise the real input, clear the session and manage the exported copy. The safest answer to “can I paste this?” comes from matching verified technical behavior with the organisation’s policy, not from convenience or branding alone.