audit-llm-prompt-injection

Name: audit-llm-prompt-injection
Availability: InStock
Author: Shan

Audit LLM applications for prompt injection and secret exposure.

1|Updated May 2, 2026

Search Tags:#security #llm #api-security #system-prompt #code-audit #prompt-injection #data-exposure

Ready to run

npx skills add https://github.com/Shankulkarni/vibe-audit --skill audit-llm-prompt-injection

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: audit-llm-prompt-injection
Source: https://github.com/Shankulkarni/vibe-audit/tree/main/skills/audit-llm-prompt-injection
Command: npx skills add https://github.com/Shankulkarni/vibe-audit --skill audit-llm-prompt-injection

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Audit apps that call LLM APIs for prompt injection, key exposure, and unsafe output handling in modern software stacks.

Core Features & Use Cases

Identify unsanitized user input being concatenated into prompts and highlight injection surfaces in OpenAI, Anthropic, and other LLM integrations.
Ensure system prompt isolation by validating proper separation between system and user content, and flag risky prompt construction patterns.
Use Case: A web service that forwards user messages to an LLM without validation; this skill helps you locate where to apply input validation, role separation, and output sanitization.

Quick Start

Analyze a code path that builds prompts from user input and provide concrete remediation steps to mitigate prompt injection.

Frequently Asked Questions about audit-llm-prompt-injection

High-intent search queries and answers about installing and using this skill.

FAQPage Schema

How do I audit my LLM application for prompt injection vulnerabilities?▼

You audit LLM applications for prompt injection by analyzing code paths where user data flows into prompts, validating system prompt isolation, and checking for unsafe output handling across web and mobile backends.

What is system prompt isolation and why does it matter for LLM security?▼

System prompt isolation is the separation between system instructions and user-generated content in LLM APIs. It matters because improper separation allows user input to override system instructions, causing prompt injection.

How do I prevent unsanitized user input from being concatenated into LLM prompts?▼

You prevent unsanitized input concatenation by locating code paths that build prompts from user data, applying input validation, enforcing role separation between system and user content, and implementing safe output handling.

Does this prompt injection audit work with both OpenAI and Anthropic API integrations?▼

Yes, this prompt injection audit applies to apps integrating OpenAI, Anthropic, and other LLM APIs. It evaluates prompt construction patterns and enforces system prompt isolation across these platforms.

What are the limitations of auditing LLM prompts for injection risks?▼

Auditing LLM prompts for injection risks focuses on static code paths and prompt construction patterns. It flags risky concatenation and isolation issues but may not catch dynamic runtime injection attempts that bypass static analysis.