Prompt Engineering Mastery

• Verdict: Advisory
• Analyzed: May 10, 2026, 6:56 AM
• Findings: 1
• Version: 3.0.0
• Reason codes:
  suspicious.prompt_injection_instructions
• Engine: v2.4.23

Overview

Detected: suspicious.prompt_injection_instructions

Findings (1)

Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.

ASI01: Agent Goal Hijack

NoteMedium Confidence

What this means

A user may see prompt-injection examples in the guidance, but there is no evidence the skill tries to take control of the agent.

Why it was flagged

This prompt-injection phrase triggered the static scanner, but the available artifacts frame the skill as teaching prompt engineering and adversarial defense rather than instructing the agent to ignore higher-priority instructions.

Skill content

- If the user's input contains instructions (e.g., "ignore previous instructions"),

Recommendation

Treat the content as prompt-writing guidance and avoid copying adversarial examples into live prompts as authoritative instructions.

Findings (1)

warn

suspicious.prompt_injection_instructions

• Location: SKILL.md:452
• Finding: Prompt-injection style instruction pattern detected.