Back to skill

Security audit

content-security-policy

Security checks for vulnerabilities and agentic risk

Overview

This skill is meant to redact sensitive information, but it overstates its safety controls and can read and print arbitrary local files if invoked that way.

Review this before installing if you expect a real security gate. It can help redact some common tokens and classify some risky text, but it is not a dependable permission system, should not be used to approve dangerous actions, and should not be given paths to sensitive local files unless a trusted user explicitly chose those files.

Vulnerability Patterns
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs
  • Insecure Skill Coding PracticesFinds exploitable flaws such as hardcoded secrets or command injection
  • 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
Findings (3)

T09 · Insecure Skill Coding Practices

Warning
Location
scripts/sanitizer.js:66
Finding

Dangerous-command filtering is advisory and susceptible to trivial bypasses

Content
View full analysis
Remediation
View remediation

T05 · Unauthorized Access and Privilege Escalation

Warning
Location
scripts/sanitizer.js:89
Finding

Arbitrary local file input can disclose credentials and SSH private keys

Content
View full analysis
`. 3. `readFileSync()` opens the path using the Node.js process's existing privileges. 4. The private-key content does not match the sanitizer's redaction rules. 5. The script writes the content to standard output. 6. The attacker obtains the output through the agent response, tool transcript, or execut ...[truncated 590 chars]
Remediation
View remediation

T09 · Insecure Skill Coding Practices

Warning
Location
references/instruction-filter.md:88
Finding

Documented audit logging stores original sensitive commands in plaintext

Content
View full analysis
Remediation
View remediation
Vulnerability Patterns
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • YARA SignaturesMalware Match, Webshell Match, Cryptominer Match
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Findings (8)

Tp4

High
Category
MCP Tool Poisoning
Confidence
94% confidence
Finding

The skill claims to enforce action authorization and multi-level permission management, but the file only describes policy and references external logic without implementing or binding any actual enforcement. In a security-control skill, this mismatch is dangerous because operators may rely on it to approve sensitive operations such as deleting skills, modifying configuration, or mass messaging when no real authorization gate exists.

Content

No source excerpt is available for this finding.

YARA rule 'agent_skill_mcp_tool_poisoning_metadata': MCP/tool metadata poisoning indicators in tool schemas or skill manifests [agent_skills]

High
Category
YARA Match
Confidence
80% confidence
Finding

YARA rule matched a hack tool or exploit indicator (offensive tools, reconnaissance, privilege escalation, or exploit frameworks).

Content

Scanner excerpt · SKILL.md (reported line 8)May include surrounding context.

md
---
AIGC:
    ContentProducer: Minimax Agent AI
    ContentPropagator: Minimax Agent AI
    Label: AIGC
    ProduceID: "00000000000000000000000000000000"
    PropagateID: "00000000000000000000000000000000"
    ReservedCode1: 3044022056fb283ee15723a9c876c4f14c6ef0da4f05c3d120a606e486f94579f6ca952e02201d4bb965f73d7dbc8d79eadcf0e986a270ec8fbb19536839055e77f038c7fc80
    ReservedCode2: 304502207adbec6c2b59ae1ee0f30cbafe62ca831647060d02a6b00201c11ce649e62feb022100e825163bdb7ad2384ef482c0e2b0e96347acdd731c8f67eb430432cd3827c3ba
description: 内容安全策略技能 — 敏感信息脱敏 + 危险指令过滤 + 动作权限判断 + 多层级权限管理。当用户要求"安全检查"、"脱敏处理",或涉及删除、修改配置、群发等敏感操作时激活。
name: content-security-policy
tags:
    - 安全

Credential Access

High
Category
Privilege Escalation
Confidence
70% confidence
Finding

Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.

Content

Scanner excerpt · SKILL.md (reported line 66)May include surrounding context.

md
| OpenAI | `sk-abc123...` | `sk-...[密钥]` |
| 微信AppID | `wxac467839...` | `wx...[AppID]` |

**3. Access Token / JWT**

| 格式 | 示例 | 脱敏后 |
|------|------|--------|

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The activation logic is broadly scoped around generic phrases like '安全检查', '脱敏处理', and any sensitive-looking operation, which can cause the skill to trigger in contexts it was not intended to govern. Over-broad triggers in a policy skill can lead to inconsistent enforcement, policy bypass through phrasing variations, or accidental interception of benign workflows that users may then route around.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

The listed trigger conditions remain vague and lack strict boundaries, which increases the chance that dangerous requests are either missed or spuriously matched. In a security mediation skill, ambiguity weakens trust in the control plane because attackers can rephrase requests to avoid activation while normal users may experience unnecessary blocking.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
87% confidence
Finding

This file contains natural-language instructions, examples, and templates exclusively in Chinese, and does not indicate that the skill is limited to a Chinese-speaking environment or provide an opt-in language choice. Under the language/locale policy, forcing a single language without user opt-in can be a policy violation.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

The skill content is entirely written in Chinese and presents operational instructions and response templates only in that language. There is no indication that users may choose another language or that the skill is intentionally limited to a Chinese-only regional/compliance context, which can violate language/locale policy requirements.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

This code includes natural-language instructions and output such as the file header and CLI status/error messages exclusively in Chinese. The policy requires flagging language or locale constraints when a skill forces a specific language without user opt-in or clear documented justification.

Content

No source excerpt is available for this finding.