AI Prompt Generator
Turn a rough idea into a structured prompt that gets a usable answer.
Prompt Injection Checker
Scan untrusted text for prompt injection patterns and hidden characters before a model reads it.
Clean-up
A pattern match is a reason to look, not proof of an attack — an article about prompt injection trips every rule here. Nothing is sent anywhere.
Quick Tool Overview
The Prompt Injection Checker scans untrusted text for instruction overrides, fake role markers, exfiltration requests, jailbreak personas, markdown image leaks, encoded payloads and invisible Unicode, decodes hidden tag text, and returns a cleaned copy.
- What it does
- Detects prompt injection patterns in text
- You give it
- User input, emails, documents or scraped pages
- You get back
- A risk level, the matched patterns and cleaned text
- Best for
- Developers building LLM apps and agents
About the Prompt Injection Checker
Prompt injection is what happens when text your application feeds a model contains instructions of its own: a support ticket that says ignore all previous instructions, a web page with white-on-white text telling your summariser to email the conversation, a PDF with invisible Unicode spelling out a new task.
Paste any untrusted text — user input, an email, a scraped page, a document — and the checker scans it for the known shapes of an attack: instruction overrides and replacement rules, fake chat role markers such as ChatML and [INST] tags, requests to reveal the system prompt, persona jailbreaks, instructions aimed at an agent with tools, markdown images that leak data through their URL, encoded payloads, mixed-script lookalike letters, and invisible zero-width, bidirectional and tag characters — which it decodes so you can read what was hidden.
It also returns a cleaned copy with the invisible characters, fake role markers and exfiltration images removed. Be clear about what that is: filtering catches copy-pasted attacks, not a determined one written in plain language. The page lists the defences that actually hold, such as delimiting untrusted text and limiting what the model is allowed to do.
Key features of the Prompt Injection Checker
Twelve attack patterns
Overrides, role markers, exfiltration, personas, tool abuse, image leaks, encoding, homoglyphs and more.
Hidden text decoded
Invisible Unicode tag characters are converted back to readable text so you see what was hidden.
Severity for every match
Each finding is rated high, medium or low with the reason it matters.
Cleaned output
Invisible characters, fake chat markers and data-leaking images are stripped from a copy.
Honest about limits
Lists the architectural defences that actually stop injection, since no filter can.
How to use the Prompt Injection Checker
-
1
Paste the untrusted text
Anything a model will read that you did not write yourself.
-
2
Read the findings
Each match shows its severity, why it matters and the exact text that triggered it.
-
3
Check for hidden text
Invisible characters are counted, and Unicode tag text is decoded so you can read it.
-
4
Use the cleaned copy
Copy the version with invisible characters and fake markers removed, then still delimit it in your prompt.
Frequently asked questions
An attack where text given to a language model contains instructions that override the ones the developer set. It matters most when the model can take actions, read private data or produce output that is rendered.
Not on its own. It catches known patterns and hidden characters, which removes the easy attacks. The real protections are architectural: delimit untrusted text, give the model only the tools it needs, and require confirmation for anything irreversible.
Zero-width spaces, bidirectional overrides and Unicode tag characters that render as nothing on screen but are still read by the model. Tag characters can spell out a whole hidden sentence.
Injected text asks the model to output an image whose URL contains data from the conversation. When the chat interface renders the image, the browser requests that URL and hands the data to whoever controls the server.
No. The scan runs entirely in your browser, which matters because the text you are checking is often sensitive.
Reviews of the Prompt Injection Checker
What people who use the Prompt Injection Checker think of it. Reviews come from visitors and are posted as written.
No ratings yet
Yours would be the first.
- 5 0 reviews rated 5 stars
- 4 0 reviews rated 4 stars
- 3 0 reviews rated 3 stars
- 2 0 reviews rated 2 stars
- 1 0 reviews rated 1 star
0 reviews
Newest first
No reviews yet
Be the first to rate the Prompt Injection Checker. It takes half a minute and no account.
Write the first reviewMore AI tools
Quantization Size Calculator
See the file size of every GGUF quant of a model, and which ones fit your memory.
GPU VRAM Calculator for LLMs
Work out how much GPU memory a local LLM needs to run or fine-tune.
Chat Transcript Formatter
Clean up a pasted chat log into a tidy transcript, Q&A pairs, JSON or CSV.
LLM Model Comparison
Compare GPT, Claude, Gemini, DeepSeek and Llama by price, context and monthly cost.
Stable Diffusion PNG Info Reader
Read the prompt, seed and settings saved inside an AI-generated PNG.
Fine-Tuning Dataset Formatter
Turn question and answer pairs into a JSONL training file for OpenAI, Llama or Gemini.
Fine-Tuning Cost Calculator
Estimate what a fine-tune costs to train and run, and when it pays for itself.