# Sanitize Untrusted Text Before It Reaches a Model

> Strip invisible carriers, neutralise instruction-like markup and fence untrusted content before you paste it into a prompt.

This tool prepares untrusted text for inclusion in a prompt: it strips invisible carriers such as Unicode tag characters and bidirectional overrides, escapes structural markup so the content cannot break out of its own delimiter, wraps it in a fence with an explicit data-not-instructions preamble, and reports any instruction-shaped phrases for review rather than silently deleting them.

**URL:** https://convertto.tech/t/prompt-injection-sanitizer
**Category:** AI & LLM Tools (https://convertto.tech/c/ai-tools)
**Privacy:** Runs entirely in the browser; no upload
**Cost:** Free, no sign-up
**Last updated:** 2026-08-01

## Key facts

- **Privacy:** Runs entirely in your browser — nothing is uploaded
- **Cost:** Free, unlimited, no sign-up

## How to use

1. Enter or paste your untrusted text.
2. Turn "Fence the content" on or off as needed.
3. Choose the fence style.
4. Turn "Neutralise structural markup" on or off as needed.
5. Press Run, then download the result when it is ready.

## FAQ

### Does this make my application injection-proof?

No, and any tool that claims otherwise is selling something. Sanitising input removes the invisible carriers and stops content escaping its delimiter, which is worth doing. It cannot stop a plainly-worded instruction, because that is indistinguishable from legitimate content. The defences that work are architectural: least privilege, confirmation on irreversible actions, and keeping credentials out of the model context.

### Why not just delete the suspicious phrases?

Because a security document, a bug report or a page about prompt injection legitimately contains those phrases, and silently editing a document you were asked to analyse is its own bug. They are reported so a human can decide.

## Sources

- [Unicode Technical Report #36 — Security Considerations](https://www.unicode.org/reports/tr36/) — Unicode Consortium
- [Unicode Technical Standard #39 — Security Mechanisms](https://www.unicode.org/reports/tr39/) — Unicode Consortium
- [OWASP Top 10 for LLM Applications](https://owasp.org/www-project-top-10-for-large-language-model-applications/) — OWASP

## Related tools

- [Invisible Character Detector](https://convertto.tech/t/invisible-character-detector): Find zero-width spaces, directional overrides, Unicode tag characters and homoglyphs hidden in pasted text — then strip them.
- [Context Window Calculator](https://convertto.tech/t/context-window-calculator): Paste a document and see which models it fits inside, how much of each window it fills, and what is left for the reply.
- [Edge Case Test String Generator](https://convertto.tech/t/edge-case-string-generator): Generate the strings that break input handling — Unicode, RTL, emoji, injection patterns and boundary lengths.
- [LLM Model Comparison](https://convertto.tech/t/llm-model-comparison): Every current model side by side: price per million tokens, context window, max output, modalities and cache rates.
- [LLM Token Counter](https://convertto.tech/t/llm-token-counter): Count tokens with the real BPE vocabulary, see every token coloured in place, and compare the count across models.
- [AI API Cost Calculator](https://convertto.tech/t/ai-api-cost-calculator): Work out what a prompt costs per call, per day and per month, and compare the same workload across every major model.
- [HTTP Security Header Analyser](https://convertto.tech/t/http-header-analyser): Paste response headers and get a security grade, with what is missing and why it matters.
- [LLM Stream Parser](https://convertto.tech/t/llm-stream-parser): Paste a raw server-sent-event stream and get the reconstructed message, tool calls, usage and stop reason.