Devil's Advocate (魔鬼代言人)
Security checks across malware telemetry and agentic risk
Overview
Prompt-injection indicators were detected in the submitted artifacts (unicode-control-chars); human review is required before treating this skill as clean.
This skill appears safe to install if you want a blunt critique assistant. Expect intentionally critical output with no praise; do not use it when you need balanced coaching, emotional support, or open-ended collaboration. ClawScan detected prompt-injection indicators (unicode-control-chars), so this skill requires review even though the model response was benign.
Publisher note
A structured critique agent that summons 5 stakeholder personas (Skeptical User, Penny-Pinching CEO, Overworked Engineer, Cautious Compliance Officer, Calm Competitor) to challenge proposals from multiple angles. Each persona delivers 3 pointed questions. Outputs tiered improvement recommendations. No praise, no flattery, just sharp critique.
VirusTotal
VirusTotal findings are pending for this skill version.
Risk analysis
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.
When active, the assistant may give sharp critique immediately instead of first checking whether the user wants a softer or more balanced review.
The skill tells the agent to begin the five-persona critique immediately rather than ask for confirmation. This changes response style, but it is consistent with the stated devil's-advocate purpose and does not involve tools or external actions.
不询问确认，直接开始 5 角色评审
Use this skill only when you want direct critical feedback; ask explicitly for balance or disable the skill if you want encouragement, brainstorming, or neutral analysis.