User documentation: TextTool#

Purpose#

TextTool is a web-based tool for iterative text editing with locally operated large language models. It is used for quick and structured revision of texts using predefined editing functions, without you having to deal with complex prompt formulations.

The basic principle: you work with a two-column interface in which you enter your source text, select an editing function and see the result directly in comparison. The transfer function allows you to continue working step by step, with all intermediate steps documented in a history.

Funktion

Range of functions#

The interface offers the following core functions:

Basic editing

  • Correct: Correction of spelling, grammar and punctuation errors while retaining the original style
  • Rephrase: Conversion to a professional, factual and academic style of language
  • Summarise: Condensing the text to the essential information and core ideas

Translation and variants

  • Translation: Direct translation into German or English while preserving tone and style
  • 3 variants: Automatic creation of three stylistic alternatives (more formal, more accessible, more understandable)

Target group adaptation

  • Gender neutrality: Revision with gender-neutral language using contemporary forms
  • Easy language: Translation into easy language according to official standards
  • Language analysis: Detailed examination of stylistic consistency, tone and target group appropriateness

Analysis and reflection

  • Content analysis: In-depth examination of key messages, argumentation structure and logical consistency
  • Discussing ideas: Critical and constructive examination of concepts presented in the text
  • Preparing tasks: Structuring texts into clear tasks with objectives, steps and resources

Advanced functions

  • Additional instructions: Supplement each editing function with individual specifications
  • Custom command: Formulate your own editing instructions
  • History management: Documentation of up to ten editing steps with the option to restore
  • Export: Output in .txt, .md, .html and .docx formats

Operation#

Basic workflow

  1. Enter your text in the left-hand area ‘ORIGINAL TEXT’
  2. Select one of the editing functions by clicking on the corresponding button
  3. The result appears in the right-hand area ‘RESULT’
  4. Check the result and click on ‘Apply’ to use it as the new source text
  5. Repeat the process for further editing steps

Using additional instructions

You can combine each editing function with additional specifications:

  1. Click on ‘Additional instruction…’
  2. Enter your additional instruction (e.g. ‘but use a more relaxed tone’).
  3. Then select an editing function.
  4. The additional instruction remains active and is taken into account for all subsequent functions until you clear the field.

Create your own commands

For completely customised edits:

  1. Click on ‘Custom command…’
  2. Formulate your instruction (e.g. ‘Add a motivating introduction’)
  3. Click on “Execute”

Use history

The history documents your editing steps:

  1. Switch to the ‘History’ tab
  2. Select an entry from the list
  3. Click on ‘View’ to preview or ‘Set as current’ to restore
  4. Click on ‘Clear history’ to reset the history

Export text

Three export options are available:

  • Export original: Saves the current original text
  • Export Result: Saves the current result
  • Export Both: Saves both texts one below the other

Each export automatically creates all four file formats at the same time.

Important notes

  • The copy functions in the text areas allow you to copy directly to the clipboard
  • For long texts, please note the processing time of the language model used
  • Session-based storage means that all data is lost when the browser is closed
  • Export important interim results regularly

Application example#

Initial situation: You have written a technical blog article that needs to be understandable to a wider audience without losing any precision.

Step 1: Paste the source text into the ‘ORIGINALTEXT’ field

Step 2: Click on “Correct” to eliminate linguistic errors

Step 3: Check the result and click on ‘Apply’ – the corrected text becomes the new source text

Step 4: Activate ‘Additional instruction…’ with the note ‘more understandable for laypeople’

Step 5: Click on ‘Rephrase’ – the result shows a more accessible version

Step 6: If the result is still too technical, switch to the ‘History’ tab and restore an earlier version

Step 7: Try again with ‘3 variants’ to see different wording options

Step 8: Export the final result in all formats by clicking on ‘Export result’

Result: An iteratively improved text with documented intermediate steps and various editing options to choose from.

Recommendations for efficient use#

  • Work iteratively: Use the Apply function for gradual refinements instead of large single steps
  • Export regularly: The history is not saved across browser sessions
  • Use additional instructions sparingly: overly complex specifications can impair the quality of the editing
  • Experiment with the 3-variant function: you will get different perspectives on the same content
  • Check the language analysis before making major rewordings: you will receive valuable tips on stylistic consistency
  • Combine functions strategically: correct before rephrasing to ensure clean source texts
  • Use the history for quality assurance: compare different processing stages directly
  • Pay attention to the output language for translations: the function automatically recognises the source language

System limitations#

The LLM text processing interface is subject to the following basic limitations:

Technical limitations

  • The maximum text length is limited by the token limit of the language model used (2000 tokens for the response by default)
  • The processing speed depends on the performance of your locally operated language model
  • The quality of the results is directly dependent on the language model used

Functional limitations

  • The tool only processes text; formatting, tables or embedded media are not transferred
  • The history is limited to ten entries and is not stored permanently
  • The system cannot integrate external sources or verify facts
  • Contextual understanding beyond the individual editing instance is not guaranteed

No support for

  • Real-time collaboration between multiple users
  • Automatic version control with Git integration
  • Direct connection to content management systems
  • Editing of tables or structured data formats
  • Creation of graphics or visual elements

User responsibility

  • Technical accuracy: The system does not check the content accuracy of the output
  • Copyright: For third-party texts, the responsibility for lawful use lies with the user
  • Data protection: Sensitive data should not be processed via the interface if the language model is not operated locally and in isolation

Summary#

The LLM text editing interface is a practical tool for structured, iterative text editing with local language models. It standardises common editing tasks through predefined functions while allowing for individual customisation.

The strength of the tool lies in its combination of user-friendliness and flexibility: you get quick access to common text operations, but retain full control through additional instructions and your own commands. The documented history allows you to compare different editing paths and return to them if necessary.

Your role as a user remains central: the interface supports you in the technical execution of text editing, but the evaluation of content, strategic selection of functions and final quality control remain your responsibility. Think of the tool as an advanced writing assistant, not an autonomous text producer.

Backward TalkToDocuments Translate Forward