招聘全流程助手

• Outcome: Review
• Latest audit: Jun 10, 2026 at 9:43 AM UTC
• Version:
  1.0.0

Overview

This recruiting assistant is purpose-aligned, but it handles sensitive candidate data and external Feishu actions with insufficient approval and privacy safeguards.

Install only if you are comfortable giving the assistant access to Feishu recruiting workflows and will enforce human approval before messages, calendar invites, table creation, candidate-data storage, or onboarding document collection. Configure Feishu tables and chats with least-privilege access, avoid broad sharing of candidate details, set retention/deletion rules, and do not use the password-disclosure checklist item.

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
• MCP Tool Poisoning: Hidden Instructions, Unicode Deception, Parameter Description Injection
• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands
• Privilege Escalation: Excessive Permissions, Sudo/Root Execution, Credential Access

Findings (9)

Intent-Code Divergence

Medium

• Confidence: 87% confidence
• Finding: The skill promises that candidate information will only be visible to recruiting-related personnel, yet elsewhere directs use of broad group messaging for internal referral promotion. In a recruiting context, even incidental disclosure of candidate-identifying details or role-specific status in wider chats can violate confidentiality expectations and create unnecessary exposure of personal data.

Vague Triggers

Medium

• Confidence: 90% confidence
• Finding: The activation triggers are broad, covering many common recruiting-related phrases without defining clear boundaries or disambiguation conditions. This can cause the skill to activate in contexts where the user did not intend recruiting automation, increasing the chance of unsolicited processing of hiring data or privacy-impacting actions.

Missing User Warnings

Medium

• Confidence: 95% confidence
• Finding: The README describes handling candidate pipelines, interview scheduling, IM notifications, and recruitment records, but does not prominently require user-facing notice, consent, or approval before privacy-impacting actions. In a recruiting context, this creates real risk of exposing personal data, contacting candidates without authorization, or processing sensitive employment information without adequate safeguards.

Missing User Warnings

Medium

• Confidence: 91% confidence
• Finding: The skill describes automatic candidate outreach, reminder sequences, and automatic record creation in tables without requiring explicit user confirmation or warning about privacy-sensitive processing. In recruitment workflows, this can lead to unsolicited messaging, over-collection, or storage of personal data without proper consent, legal basis, or operator awareness.

Missing User Warnings

Medium

• Confidence: 95% confidence
• Finding: The checklist directs HR to collect multiple categories of highly sensitive personal data, including identity documents, household registration pages, bank details, medical reports, and scanned copies for storage, but gives no guidance on data minimization, lawful basis, secure storage, access control, retention, or redaction. In a recruiting/onboarding skill, this omission can directly normalize unsafe handling of personal information and increase the risk of privacy breaches, insider misuse, or noncompliance with data protection obligations.

Missing User Warnings

Medium

• Confidence: 98% confidence
• Finding: The onboarding steps include '账号密码告知' (telling the new hire the account password), which encourages insecure credential distribution and potentially shared or pre-set passwords. In an HR onboarding process, this is especially risky because it can lead to credential leakage, weak auditability, account takeover, and persistence of compromised initial credentials if secure first-login reset controls are absent.

Missing User Warnings

Medium

• Confidence: 91% confidence
• Finding: The README describes automatically writing candidate tags, match scores, and other recruiting data into Feishu tables without any mention of privacy notice, lawful basis, retention limits, access control, or handling of personal information. In a recruiting workflow, this can lead to unauthorized collection, over-sharing, and non-compliant processing of sensitive candidate data, especially given the explicit storage of profiles, labels, and evaluation outcomes.

Missing User Warnings

Medium

• Confidence: 94% confidence
• Finding: The documentation states the AI will create Feishu calendar events, invite interviewers and candidates, and send interview instructions, but it does not clearly warn users that the skill may take external actions on their behalf. In a recruiting context, these actions can expose candidate identities, schedules, and meeting details or trigger unintended communications if the user did not expect autonomous execution.

Missing User Warnings

Medium

• Confidence: 95% confidence
• Finding: The documentation says the AI will create recruiting reports and Feishu multidimensional tables containing HC status, channel effectiveness, and candidate pipeline data, but it does not warn that these tables may contain sensitive hiring and personal information. In this context, silent creation or broad sharing of such data structures increases the risk of privacy breaches, internal information exposure, and unauthorized access to staffing plans or candidate records.

VirusTotal

65/65 vendors flagged this skill as clean.