Back to skill

Security audit

Eastern Philosophy Methodology东方哲学方法skill技能包

Security checks for vulnerabilities and agentic risk

Overview

ClawScan rates this skill suspicious because scanners report privacy and confidentiality risks, manipulative guidance, and incomplete analysis of two referenced registries; this does not establish malicious intent.

The scanner recommends against installing, and the overall verdict is suspicious. Consider avoiding installation unless you can review the full artifact and accept the reported risks; analysis was partial, so the scan cannot resolve every concern.

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
Vulnerability Patterns
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Output HandlingUnvalidated Output Injection, Cross-Context Output, Unbounded Output
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (18)

Incomplete referenced artifact analysis

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 24)May include surrounding context.

md
1. **匹配场景(含场景准入前检)**:查 `application-skills/_registry.md`,命中候选后**必须先做场景准入前检**——对照注册表"适用域/排除"列与该应用 Skill 的"场景定义"四要素(目标/约束/角色/典型输入)逐项核验:约束性前提不成立即视为**未命中**,转查其他应用 S

Incomplete referenced artifact analysis

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 25)May include surrounding context.

md
2. **未命中**:查 `atomic-skills/_registry.md` 直接取用原子(遵循其适用条件与边界);确有必要沉淀为新应用 Skill 的,走进化流水线(`agents.md` 第 8 节)的 app-add 提案。

Pre-interview background checks lack candidate-consent and confidentiality warnings.

High
Category
Output Handling
Confidence
96% confidence
Finding

The workflow directs contacting multiple third parties before interviewing a candidate without requiring candidate notification, an appropriate legal basis, or confidentiality controls. This can reveal a confidential job search and collect consequential personal allegations; the later recommendation to prevent candidates from learning about inquiries increases the concern.

Content

No source excerpt is available for this finding.

Mandatory feedback logging lacks a warning about persisting potentially sensitive user information.

Medium
Category
Output Handling
Confidence
88% confidence
Finding

The workflow requires writing feedback to persistent files after every use without a sensitivity check or user choice. In negotiation, workplace, and organizational scenarios, this can unnecessarily retain confidential user information in a workspace that may later be shared or committed.

Content

No source excerpt is available for this finding.

The feedback-record specification omits privacy safeguards for stored scenarios and execution details.

Medium
Category
Output Handling
Confidence
90% confidence
Finding

The feedback specification solicits scenarios, execution details, and actual outcomes for persistent storage without minimization or confidentiality requirements. Such records can preserve sensitive business decisions or identifiable third-party information beyond the original interaction.

Content

No source excerpt is available for this finding.

Instructions to maintain a behavioral dossier on a manager omit a privacy warning.

Medium
Category
Output Handling
Confidence
94% confidence
Finding

The skill explicitly directs maintaining and repeatedly updating an identifiable manager's behavioral dossier, including inferred anxieties and boundaries. This goes beyond preparing a meeting and can create sensitive, speculative personnel records without collection or retention limits.

Content

No source excerpt is available for this finding.

The recruitment workflow requires background checks without warning about candidate privacy.

Medium
Category
Output Handling
Confidence
87% confidence
Finding

This source independently makes background checking a required recruitment step without an authorization or confidentiality gate. Following it literally can disclose candidate information to third parties even when urgent team formation does not justify that disclosure.

Content

No source excerpt is available for this finding.

The workflow directs ongoing profiling of customer contacts without a privacy warning.

Medium
Category
Output Handling
Confidence
95% confidence
Finding

The workflow explicitly calls for continuously recording individual customer contacts' priorities and inferring hidden motivations, alongside probing emotional vulnerability. Without boundaries, the resulting sales dossiers can contain unnecessary personal assessments rather than legitimate account requirements.

Content

No source excerpt is available for this finding.

The workflow instructs sharing forecast plans with suppliers without warning about disclosure of confidential business data.

Medium
Category
Output Handling
Confidence
88% confidence
Finding

This step explicitly directs sharing forecast plans with an external supplier, rather than merely describing an available capability. Such plans can reveal nonpublic demand, launch timing, or procurement strategy, but the skill provides no confidentiality warning or disclosure limits.

Content

No source excerpt is available for this finding.

Leak-tracing instructions omit warnings about exposing real secrets and tracking identifiable recipients.

Medium
Category
Output Handling
Confidence
91% confidence
Finding

The workflow directs distributing different information versions to trace leaks and repurposing a suspected adversary's channel. It does not warn that these actions could further disclose confidential information or associate individuals with suspected misconduct; the later instruction to investigate facts does not address these exposure and privacy risks.

Content

No source excerpt is available for this finding.

L70 instructs publicizing the canceled-project list without a confidentiality warning.

Medium
Category
Output Handling
Confidence
88% confidence
Finding

The workflow directs that canceled projects be listed and made public, without specifying an authorized audience. Such a list can expose confidential business priorities or project information, but the skill provides no warning or disclosure review requirement.

Content

No source excerpt is available for this finding.

The mediation workflow normalizes covertly exploiting disputes to consolidate the mediator’s control.

Medium
Category
Prompt Injection
Confidence
94% confidence
Finding

The operational workflow tells a purported mediator to let powerful parties fight and selectively praise winners to consolidate personal control. This is deliberate manipulation of parties relying on an intermediary, not merely historical discussion, and can escalate disputes while concealing a conflict of interest.

Content

No source excerpt is available for this finding.

L81 directs full public disclosure of mediation facts without protecting confidential party information.

Medium
Category
Output Handling
Confidence
95% confidence
Finding

The fallback explicitly directs comprehensive public disclosure of facts and rules to restore the mediator's credibility. Mediation facts can include confidential concessions, commercial information, and personal allegations, and protecting the mediator's reputation does not authorize publishing them.

Content

No source excerpt is available for this finding.

The user-research workflow lacks privacy warnings for behavioral observation and profiling.

Medium
Category
Output Handling
Confidence
87% confidence
Finding

These steps instruct observing usage behavior to infer hidden motivations, probing users at emotional extremes, and classifying users by their characteristics and desires. The skill does not warn that this involves potentially sensitive behavioral information or specify consent, authorized data access, or limits on individual profiling.

Content

No source excerpt is available for this finding.

The workflow requires public disclosure of errors and failures without warning about personal or confidential information.

Medium
Category
Output Handling
Confidence
87% confidence
Finding

The workflow explicitly calls for thoroughly exposing failures and conducting open retrospectives without defining the audience or excluding sensitive information. In an organizational knowledge repository, literal implementation could expose employee details, customer information, or confidential incident data beyond authorized readers; 'open' does not necessarily mean publication to the internet.

Content

No source excerpt is available for this finding.

The workflow calls for public launch-problem retrospectives without confidentiality or privacy warnings.

Medium
Category
Output Handling
Confidence
81% confidence
Finding

The launch workflow directs open disclosure and review of launch problems without establishing confidentiality boundaries. Such reviews can contain customer complaints, unreleased product information, partner terms, or security defects, so unrestricted distribution could disclose sensitive material even if the intended audience is internal.

Content

No source excerpt is available for this finding.

Long-term preference profiling is prescribed without a privacy warning.

Medium
Category
Output Handling
Confidence
88% confidence
Finding

The skill directs observation of a person's interests, spending, and time use, and explicitly includes building long-term preference dossiers. This can involve collecting and retaining personal information, but the boundary conditions address inference accuracy rather than privacy, consent, or appropriate data sources.

Content

No source excerpt is available for this finding.

The background-check workflow lacks a warning about candidate privacy and third-party contact.

Medium
Category
Output Handling
Confidence
86% confidence
Finding

The proposal places background checks and third-party interviews before the candidate interview without requiring candidate authorization or restrictions on contacting a current employer. This could reveal a confidential job search or collect unnecessary personal information; the compliance limitation elsewhere applies specifically to emotional probing and does not clearly govern background checks.

Content

No source excerpt is available for this finding.