other
Warning
- Location
references/03-接触策略设计.md:14- Finding
Deceptive Professional-Identity Pretexting in M&A Outreach Workflows
- Content
View full analysis
- Remediation
View remediation
Security audit
胡田 OPC导师 并购对接
Security checks for vulnerabilities and agentic risk
Overview
This is a content-only M&A outreach skill, but it repeatedly encourages pretext-based contact strategies that users should review carefully before installing.
Install only if you are comfortable reviewing and editing the outreach playbooks for compliance. Users should require truthful role, affiliation, and purpose disclosure, use only lawful public or authorized business contact data, avoid personal-data accumulation, honor opt-outs, and get legal/compliance review before contacting public-company personnel about strategic transactions.
Vulnerability Patterns
- Skill Instruction HijackingAlters the agent's session goals or safety constraints when the skill loads
- Agent Memory PoisoningWrites attacker-controlled rules into memory that affect later sessions
- Remote Payload Retrieval and ExecutionFetches external code whose behavior can change after review
- Embedded Malicious CodeShips malicious scripts inside the skill and executes them locally
- Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs
Findings (1)
Vulnerability Patterns
- Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
- Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
- Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
- MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
- Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (22)
Description-Behavior Mismatch
Medium
- Category
- Not specified by scanner
- Confidence
- 93% confidence
- Finding
The manifest description at L003 limits the skill to key-contact identification, contact strategy, and negotiation path planning. However, the documented outputs and process include '尽调配合', '交易谈判', and '签约交割' in the six-stage model (L703-L722), which are execution-phase activities beyond connection/planning and overlap with downstream M&A execution work.
- Content
Natural-Language Policy Violations
Medium
- Category
- Not specified by scanner
- Confidence
- 88% confidence
- Finding
The manifest description at L003 is Chinese-only, and the rest of the skill likewise appears to assume Chinese as the operating language. Under the stated policy, forcing a specific language without user opt-in can be a natural-language policy violation when no alternative or choice is provided.
- Content
Intent-Code Divergence
Medium
- Category
- Not specified by scanner
- Confidence
- 95% confidence
- Finding
At L022, the file states the follow-on skill is '并购执行（法律尽调）', implying legal due diligence belongs to the next stage. Yet this same skill later defines stage 4 as '尽调配合' and continues through negotiation and closing (L703-L722), directly contradicting the stated stage boundary and intent of the downstream handoff.
- Content
Ssd 1
Medium
- Category
- Not specified by scanner
- Confidence
- 97% confidence
- Finding
The skill recommends adopting benign-seeming personas such as investor or industry researcher as a tactic to gain access, rather than honestly stating M&A intent. That is a classic social-engineering pattern: it encourages misrepresentation to elicit engagement from gatekeepers and can lead to reputational harm, policy breaches, and unlawful or unethical procurement of sensitive business information.
- Content
Ssd 4
Medium
- Category
- Not specified by scanner
- Confidence
- 95% confidence
- Finding
The skill explicitly recommends trust-building, indirect positioning, and avoiding direct disclosure of acquisition intent to get past gatekeeping. In this context, that normalizes deceptive social-engineering behavior against corporate staff, increasing the risk of manipulative access, unauthorized disclosure, and compliance violations during sensitive transaction discussions.
- Content
Missing User Warnings
Medium
- Category
- Not specified by scanner
- Confidence
- 89% confidence
- Finding
The document instructs users to build detailed profiles of named individuals, including education, career history, social relationships, communication style, response habits, and intermediary access, but provides no privacy, consent, lawful-basis, or data-minimization guidance. In an M&A outreach context, this materially increases the risk of unauthorized personal data collection, intrusive profiling, and misuse of relationship intelligence for targeted influence or circumvention of normal corporate channels.
- Content
Missing User Warnings
Medium
- Category
- Not specified by scanner
- Confidence
- 93% confidence
- Finding
The template explicitly instructs users to collect, store, and track named individuals' phone numbers, email addresses, relationship information, contact history, and follow-up plans, but it does not require a lawful basis, retention limit, access controls, or deletion process. In a deal-origination context, this creates a realistic risk of unnecessary personal data accumulation, privacy violations, and misuse of contact intelligence beyond the original purpose.
- Content
Intent-Code Divergence
Medium
- Category
- Not specified by scanner
- Confidence
- 94% confidence
- Finding
The document claims a compliance-first approach, but later sections recommend approaching targets under selectively framed identities and avoiding explicit disclosure of M&A intent, especially phrases like '不提并购' and '需要绕道'. This mismatch can enable deceptive social-engineering-style outreach that gains access under false pretenses, increasing legal, ethical, and reputational risk.
- Content
Ssd 4
Medium
- Category
- Not specified by scanner
- Confidence
- 84% confidence
- Finding
The '价值交换' sequence advises identifying the target's needs, providing value, building reciprocity, and only then raising the real request. While relationship-building is normal, this framing becomes risky when used to withhold the underlying M&A objective until trust is established, creating manipulative access-gaining behavior.
- Content
Ssd 4
Medium
- Category
- Not specified by scanner
- Confidence
- 91% confidence
- Finding
The role-specific guidance for 董秘 explicitly says not to directly mention M&A because it will be filtered and that one should '绕道', which is effectively advice to bypass screening and conceal the true purpose of contact. In the context of an M&A contact strategy skill, this is more dangerous because the entire workflow is about gaining sensitive access to decision-makers, making the concealment operational rather than incidental.
- Content
Ssd 4
Medium
- Category
- Not specified by scanner
- Confidence
- 92% confidence
- Finding
The risk-control advice says to '先建立关系，再逐步深入' in order to avoid exposing intent too early, which directly endorses delaying disclosure until access is secured. That pattern is characteristic of deceptive pretexting and is especially concerning in a business-development/M&A targeting context where contact recipients may make decisions based on incomplete or misleading information.
- Content
Missing User Warnings
Medium
- Category
- Not specified by scanner
- Confidence
- 95% confidence
- Finding
The due diligence section encourages providing materials, arranging interviews, and sharing company information, but lacks clear limits on what sensitive information may be disclosed and under what controls. In M&A workflows, this can expose confidential business, employee, financial, customer, or trade secret information before appropriate scoping, redaction, need-to-know review, and secure sharing mechanisms are in place.
- Content
Missing User Warnings
Medium
- Category
- Not specified by scanner
- Confidence
- 94% confidence
- Finding
The skill explicitly asks for and structures collection of key-person contact details without any accompanying requirement to verify lawful source, consent, or permitted use. In an M&A outreach context, this can normalize scraping, unauthorized use of personal data, and non-compliant contact practices, creating privacy, harassment, and regulatory risk.
- Content
Missing User Warnings
Medium
- Category
- Not specified by scanner
- Confidence
- 96% confidence
- Finding
The output schema directs the skill to generate contact-acquisition plans and outreach scripts but omits safeguards against invasive, deceptive, or repeated unsolicited contact. Because the skill is specifically designed to identify and approach individuals in corporate transactions, missing compliance warnings makes misuse more likely and increases the chance of privacy violations or improper solicitation.
- Content
Missing User Warnings
Medium
- Category
- Not specified by scanner
- Confidence
- 93% confidence
- Finding
The guide explicitly promotes identifying key decision-makers and obtaining their contact details, but it does not pair that capability with privacy, consent, data minimization, or lawful-source requirements. In an M&A outreach context, users may infer that collecting and using personal or semi-personal contact data for unsolicited approaches is endorsed, increasing the risk of privacy violations, misuse of personal data, and non-compliant outreach.
- Content
Missing User Warnings
Medium
- Category
- Not specified by scanner
- Confidence
- 90% confidence
- Finding
The quick-start templates and scenarios normalize repeated outreach, batch contact planning, and follow-up scripting without warning about anti-spam, harassment, or contact-frequency limits. Because the skill is designed to operationalize external contact campaigns, omission of compliance guidance makes abusive or non-compliant mass outreach easier to scale.
- Content
Natural-Language Policy Violations
Medium
- Category
- Not specified by scanner
- Confidence
- 92% confidence
- Finding
This Python file contains user-facing natural-language content entirely in Chinese, including the module description and generated templates. Under the policy, forcing a specific language without user opt-in is a natural-language locale violation unless the constraint is explicitly documented and justified.
- Content
Natural-Language Policy Violations
Low
- Category
- Not specified by scanner
- Confidence
- 90% confidence
- Finding
The file presents all operational guidance exclusively in Chinese and does not indicate that the user can choose another language. Under the stated policy, forcing a specific language without user opt-in can be a natural-language policy violation unless the locale limitation is explicitly justified.
- Content
Natural-Language Policy Violations
Low
- Category
- Not specified by scanner
- Confidence
- 77% confidence
- Finding
该文件全文仅使用中文撰写，未说明这是面向特定中文用户群体的区域性资料，也未提供可选语言或用户选择机制。根据规则，强制特定语言且无用户 opt-in 可能属于自然语言层面的组织政策违规。
- Content
Natural-Language Policy Violations
Low
- Category
- Not specified by scanner
- Confidence
- 73% confidence
- Finding
自然语言内容全程固定为中文，且未说明这是面向特定中文用户群或地区场景的限定文档。按规则，强制单一语言而无用户选择或明确合理说明，可能构成语言/区域策略层面的自然语言政策问题。
- Content
Missing User Warnings
Low
- Category
- Not specified by scanner
- Confidence
- 93% confidence
- Finding
The document includes templates to collect and store named contacts and their contact details, but it does not instruct users to obtain authorization, minimize personal data collected, or comply with privacy and data protection requirements. In an M&A outreach context, this can lead to unnecessary retention or misuse of personal information about executives and employees, creating legal, compliance, and confidentiality risk.
- Content
Natural-Language Policy Violations
Low
- Category
- Not specified by scanner
- Confidence
- 73% confidence
- Finding
文件内容、示例输入输出和模板均固定为中文，未看到任何允许用户选择语言或说明该技能仅面向特定中文场景的提示。若组织要求避免未经用户同意强制特定语言，这种默认单一语言表述可能构成自然语言层面的语言/地区策略问题。
- Content