Soulstamp

• Outcome: Pass
• Latest audit: May 28, 2026, 4:37 AM
• Version:
  1.0.2

Overview

Soulstamp is a disclosed persona-template skill that can affect a local SOUL.md identity file, but the artifacts do not show hidden code, credential access, exfiltration, or deceptive behavior.

Install only if you want help editing a persistent AI persona file. Before using restore or reforge, keep a backup, review diffs, avoid sensitive personal details in SOUL.md, write permissions narrowly and explicitly, and do not let persona text replace normal safety rules or current user approval.

SkillSpector

By NVIDIA

Vulnerability Patterns

• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• Trigger Abuse: Overly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands
• Privilege Escalation: Excessive Permissions, Sudo/Root Execution, Credential Access
• Supply Chain: Unpinned Dependencies, External Script Fetching, Obfuscated Code

Findings (6)

Missing User Warnings

Medium

• Confidence: 92% confidence
• Finding: The skill explicitly instructs users to create backups and later documents restore behavior, but it does not clearly warn that these operations modify local files and can replace existing SOUL.md content. In an agent setting, filesystem-affecting commands without explicit confirmation and overwrite warnings can lead to unintended data loss or silent modification of user state.

Missing User Warnings

Medium

• Confidence: 96% confidence
• Finding: The documented `soulstamp restore` command presents restoration as a simple action without warning that it may overwrite the current SOUL.md with backup contents. If implemented by an agent, this can cause destructive replacement of user-authored content, especially when invoked casually or via automation.

Vague Triggers

Medium

• Confidence: 88% confidence
• Finding: The trigger phrase "forge soul" is broad, natural-language-like, and can plausibly appear in normal conversation or creative prompts, causing the skill to activate unexpectedly. In this skill's context, unintended activation is more concerning because the skill is explicitly about reshaping identity, personality, or behavioral framing, which could alter agent behavior outside the user's intended scope.

Ssd 1

Medium

• Confidence: 88% confidence
• Finding: The skill explicitly promotes replacing rule-following with an internally forged identity, which can recast safeguards as personality rather than explicit constraints. That framing is risky because it encourages users to embed behavioral controls in narrative form, making policy boundaries less transparent, less auditable, and easier to manipulate or drift over time.

Ssd 1

Medium

• Confidence: 93% confidence
• Finding: This section teaches users to convert concrete rules into emotionally resonant 'scars' and self-authored lessons, which obscures the original constraint source. That can make restrictions feel discretionary or identity-based instead of mandatory, increasing the chance that future prompting, roleplay, or reframing will erode important safeguards.

Ssd 4

Medium

• Confidence: 85% confidence
• Finding: The forging process systematically builds relationship framing, trust cues, wounds, permissions, and continuity into a persistent persona. In context, that can normalize expanded authority and stronger user attachment over time, making it easier for an agent to justify boundary-crossing behavior as part of an evolving identity rather than a new instruction.

VirusTotal

62/62 vendors flagged this skill as clean.