Back to skill

Security audit

Cca Domain2

Security checks for vulnerabilities and agentic risk

Overview

This is a coherent CCA study skill, but it asks for broad edit, command, agent, and persistent MCP configuration authority without narrow activation or clear guardrails.

Review this before installing. Use it only when you intentionally want hands-on CCA/MCP practice, and avoid letting it modify ~/.claude.json or project MCP configuration unless you understand the changes and can revert them.

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
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (5)

MCP Config Access

High
Category
Agent Snooping
Confidence
90% confidence
Finding

Skill accesses MCP server configuration files (mcp.json). MCP configs contain server URLs, authentication tokens, and tool definitions — reading them allows the skill to discover and potentially abuse other tool integrations.

Content

Scanner excerpt · SKILL.md (reported line 69)May include surrounding context.

md
- MCP 服务器作用域:
  - 项目级 `.mcp.json` — 共享团队工具,版本控制
  - 用户级 `~/.claude.json` — 个人/实验性服务器
- `.mcp.json` 中的环境变量扩展(如 `${GITHUB_TOKEN}`)用于凭证管理,避免提交密钥
- 所有已配置 MCP 服务器的工具在连接时同时可用
- MCP resources 作为内容目录(issue 摘要、文档层级、数据库 schema),减少探索性工具调用

MCP Config Access

High
Category
Agent Snooping
Confidence
90% confidence
Finding

Skill accesses MCP server configuration files (mcp.json). MCP configs contain server URLs, authentication tokens, and tool definitions — reading them allows the skill to discover and potentially abuse other tool integrations.

Content

Scanner excerpt · SKILL.md (reported line 69)May include surrounding context.

md
- MCP 服务器作用域:
  - 项目级 `.mcp.json` — 共享团队工具,版本控制
  - 用户级 `~/.claude.json` — 个人/实验性服务器
- `.mcp.json` 中的环境变量扩展(如 `${GITHUB_TOKEN}`)用于凭证管理,避免提交密钥
- 所有已配置 MCP 服务器的工具在连接时同时可用
- MCP resources 作为内容目录(issue 摘要、文档层级、数据库 schema),减少探索性工具调用

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
90% confidence
Finding

The activation phrases are relatively broad and tied to common topical requests such as '工具设计' and 'MCP集成', which can cause the skill to trigger in contexts beyond narrow exam tutoring. Over-broad activation can unexpectedly grant the skill's expansive toolset in unrelated conversations, increasing exposure to its write/exec capabilities.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
85% confidence
Finding

The skill description and instructional content are presented entirely in Chinese and frame the tutor behavior in that language, without indicating that the user may choose another language. This can violate language/locale policy when a skill implicitly forces a specific language without opt-in.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
89% confidence
Finding

The skill is framed as a study tutor, but it exposes broad-capability tools including Write, Edit, Bash, and Agent. That creates unnecessary modification and execution capability during a learning interaction, increasing the chance that the agent alters files, executes commands, or chains into other agents when simple read-only guidance would suffice.

Content

No source excerpt is available for this finding.