MindShield - AI Critical Thinking & Prompting von Frog1230
Protect your critical thinking from AI overreliance. MindShield is an open-source browser extension that adds a brief reflection step before simple AI prompts, helping you think before asking AI.
Keine NutzerKeine Nutzer
Metadaten zur Erweiterung
Screenshots
Über diese Erweiterung
MindShield is a privacy-first AI prompting browser extension designed to help people use AI without automatically outsourcing their thinking. It encourages more intentional AI use while helping users remain involved in their own reasoning, problem-solving, and decision-making.
Instead of trying to stop you from using AI, MindShield identifies prompts that appear to ask AI to perform simple tasks you could potentially reason through yourself. When it detects one, it introduces a brief moment of friction that encourages you to think before automatically submitting the prompt.
MindShield evaluates your AI prompt locally on your device before it is submitted. Depending on the type of request, a prompt may pass normally, receive a soft warning, or trigger a brief reflection step.
More complex prompts — such as research, explanations, analysis, critiques, and other tasks where AI assistance can be useful — can pass through normally.
Why MindShield?
AI assistants are incredibly useful, but their convenience can make it easy to automatically rely on AI for small problems we could solve ourselves.
MindShield is an experiment in adding intentional friction to AI prompting.
The goal isn't to use less AI.
The goal is to use AI when it helps — rather than automatically using it for everything.
Privacy-first and open source
MindShield is designed around local processing.
MindShield is built for people who want to keep AI as a tool while staying actively involved in their own thinking.
Supported AI assistants
MindShield is designed to work with popular AI chat interfaces, including ChatGPT, Claude, Google Gemini, and Grok.
Learn more
GitHub: https://github.com/fluryjanis/Mind-Shield
Instead of trying to stop you from using AI, MindShield identifies prompts that appear to ask AI to perform simple tasks you could potentially reason through yourself. When it detects one, it introduces a brief moment of friction that encourages you to think before automatically submitting the prompt.
MindShield evaluates your AI prompt locally on your device before it is submitted. Depending on the type of request, a prompt may pass normally, receive a soft warning, or trigger a brief reflection step.
More complex prompts — such as research, explanations, analysis, critiques, and other tasks where AI assistance can be useful — can pass through normally.
Why MindShield?
AI assistants are incredibly useful, but their convenience can make it easy to automatically rely on AI for small problems we could solve ourselves.
MindShield is an experiment in adding intentional friction to AI prompting.
The goal isn't to use less AI.
The goal is to use AI when it helps — rather than automatically using it for everything.
Privacy-first and open source
MindShield is designed around local processing.
- Prompt evaluation happens locally on your device.
- Your prompts are not sent to an external classification server.
- No account is required.
- The project is open source and available for inspection on GitHub.
MindShield is built for people who want to keep AI as a tool while staying actively involved in their own thinking.
Supported AI assistants
MindShield is designed to work with popular AI chat interfaces, including ChatGPT, Claude, Google Gemini, and Grok.
Learn more
GitHub: https://github.com/fluryjanis/Mind-Shield
Bewertet mit 0 von 0 Bewertern
Berechtigungen und Daten
Benötigte Berechtigungen:
- Auf Browsertabs zugreifen
- Auf Ihre Daten für chatgpt.com zugreifen
- Auf Ihre Daten für chat.openai.com zugreifen
- Auf Ihre Daten für claude.ai zugreifen
- Auf Ihre Daten für gemini.google.com zugreifen
- Auf Ihre Daten für x.com zugreifen
- Auf Ihre Daten für grok.com zugreifen
Optionale Berechtigungen:
- Auf Ihre Daten für Websites in der Domain huggingface.co zugreifen
- Auf Ihre Daten für Websites in der Domain hf.co zugreifen
- Auf Ihre Daten für huggingface.co zugreifen
- Auf Ihre Daten für cdn.jsdelivr.net zugreifen
Datenerfassung:
- Der Entwickler sagt, dass diese Erweiterung keine Datenerhebung benötigt.
Weitere Informationen
- Add-on-Links
- Version
- 1.2.0
- Größe
- 25,62 MB
- Zuletzt aktualisiert
- vor einem Monat (28. Aug. 2026)
- Verwandte Kategorien
- Lizenz
- Alle Rechte vorbehalten
- Versionsgeschichte
- Zur Sammlung hinzufügen