DAF Intent Guard — 意图漂移检测

• Outcome: Pass
• Latest audit: May 29, 2026 at 8:28 PM UTC
• Version:
  1.0.0

Overview

This skill is a local helper that advises whether an agent should continue, replan, or stop when a user changes an ongoing task.

Install this if you want an agent to consult a local heuristic before continuing changed multi-step work. Treat its patch, replan, and abort outputs as advisory, and keep explicit user confirmation for important restarts or abandoned tasks.

SkillSpector

By NVIDIA

Vulnerability Patterns

• Trigger Abuse: Overly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands
• 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 (1)

Vague Triggers

Medium

• Confidence: 89% confidence
• Finding: The README states that the skill should be auto-triggered from AGENTS.md but does not define concrete trigger conditions, guardrails, or examples. In an agentic system, ambiguous automatic invocation can cause the skill to run too broadly or at the wrong time, leading to unintended task interruption, rollback, or control-flow changes during multi-step execution.

VirusTotal

66/66 vendors flagged this skill as clean.