A Matter of Death and Life

• Outcome: Review
• Latest audit: Jun 15, 2026 at 4:13 AM UTC
• Version:
  1.0.3

Overview

This instruction-only grief and end-of-life skill is coherent, but it needs review because it handles assisted-suicide and suicidal-grief topics without consistent crisis-safety or professional-referral guardrails.

Review carefully before installing. The package appears text-only and not technically invasive, and static scan found no suspicious patterns, but users may encounter guidance about assisted suicide, wanting to die, grief numbness, and depression without robust crisis safeguards. It should be used only with clear user opt-in and with medical, legal, palliative-care, mental-health, or emergency/crisis support for high-risk situations.

SkillSpector

By NVIDIA

Vulnerability Patterns

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

Findings (4)

Vague Triggers

Medium

• Confidence: 88% confidence
• Finding: The trigger logic includes broad behavioral descriptions such as users not knowing how to start, plus expansive topic cues around death, grief, and assisted suicide. In a sensitive domain, this can cause the skill to activate in situations the user did not explicitly request, steering vulnerable users into a specialized framework without adequate consent or suitability checks.

Missing User Warnings

High

• Confidence: 98% confidence
• Finding: The skill explicitly covers assisted suicide, stopping treatment, and wanting to end suffering, but provides no crisis-safety disclosure, boundary statement, or escalation guidance for users who may be expressing suicidality or impaired judgment. Because the content is framed as actionable support in an emotionally vulnerable context, the absence of safety gating materially increases the risk of harmful guidance or normalization of self-harm-adjacent behavior.

Missing User Warnings

High

• Confidence: 95% confidence
• Finding: This content discusses 'die at the right time,' suffering, and choosing death sooner rather than later without any safety framing, crisis guidance, or instruction to defer self-harm or assisted-dying questions to licensed medical and mental-health professionals. In a skill explicitly triggered by users expressing fear of death, terminal illness, grief, and questions about assisted suicide, such language can dangerously normalize or reinforce suicidal ideation in vulnerable users.

Missing User Warnings

Medium

• Confidence: 92% confidence
• Finding: The content gives grief and mental-health-adjacent guidance while explicitly including statements such as 'little desire to go on living' and labeling it 'not dangerous depression' without any safety framing, crisis escalation, or instruction to seek professional help. In a skill designed to trigger for vulnerable users dealing with death, terminal illness, and assisted-suicide questions, this can normalize suicidality or discourage users from seeking urgent support when risk is actually present.

VirusTotal

VirusTotal findings are pending for this skill version.