法务助手（Legal Assistant）

• Outcome: Pass
• Latest audit: May 28, 2026 at 11:29 AM UTC
• Version:
  0.1.0

Overview

This is a disclosed legal workflow assistant with sensitive but expected contract-review and ledger-support functions, and no executable or hidden behavior found.

Install only if you intend to use this with contract and legal-operation data. Limit Feishu, Hermes Agent, knowledge-base, archive, and ledger permissions to authorized users; require human review before binding legal advice, external replies, final approvals, ledger writes, or seal/signature workflows; and use test data or desensitized examples outside production.

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 (3)

Vague Triggers

Medium

• Confidence: 91% confidence
• Finding: The skill advertises broad end-to-end legal capabilities, including review, advice, approval-draft generation, and ledger maintenance, without clear activation constraints, permission checks, or explicit negative examples of disallowed use. In a legal workflow, this can cause users or downstream agents to over-trust the skill and invoke it for high-risk actions beyond safe advisory boundaries, increasing the chance of unauthorized legal decision support or operational side effects.

Missing User Warnings

Medium

• Confidence: 95% confidence
• Finding: The description says the skill completes a full contract/legal closed loop including contract ledger updates, but it does not prominently warn users that records may be modified or reminder tasks created. Because contract ledgers are systems of record, ambiguous automation can lead to unauthorized data changes, incorrect compliance tracking, or workflow actions taken without informed user consent.

Missing User Warnings

Medium

• Confidence: 97% confidence
• Finding: The workflow explicitly states that contract ledger data is automatically extracted and entered after prescreening or circulation, but it does not define approval gates, confirmation prompts, or safeguards against writing incorrect or sensitive data. In a legal operations context, automatic entry can corrupt contractual records, trigger erroneous reminders, or create compliance and audit issues if the extracted fields are wrong or the action is unauthorized.

VirusTotal

65/65 vendors flagged this skill as clean.