fuzailkhannn

Prompt Injection Audit

Name: Prompt Injection Audit
Author: fuzailkhannn

Security community

SOURCE https://github.com/fuzailkhannn/Prompt-Injection-Audit/blob/main/commands/prompt-injection-audit.md

Description

--- description: Audit an LLM-backed feature for prompt-injection exposure. Read-only, findings only. argument-hint: "[file or directory to audit]" --- Use the `prompt-injection-audit` skill to audit the target below. **Target:** $ARGUMENTS If the target is empty, find the LLM-backed code in the current working directory and audit that, saying up front which files you picked and why. Hold the skill's contract exactly, since these are the points an audit usually drifts on: - Read `references

Installation

Installs to ~/.claude/skills/fuzailkhannn-prompt-injection-audit-prompt-injection-audit/SKILL.md

Terminal

mkdir -p ~/.claude/skills/fuzailkhannn-prompt-injection-audit-prompt-injection-audit && curl -fsSL https://raw.githubusercontent.com/fuzailkhannn/Prompt-Injection-Audit/HEAD/commands/prompt-injection-audit.md -o ~/.claude/skills/fuzailkhannn-prompt-injection-audit-prompt-injection-audit/SKILL.md

Restart Claude Code, or start a new session, for it to be picked up.