Back to skill

Security audit

CLAUDE.md / AGENTS.md 体检器

Security checks for vulnerabilities and agentic risk

Overview

ClawScan marks this skill benign with medium overall confidence: it found no AIG issues, while Skillspector flagged unpinned Docker image references, mostly as examples or documentation, and its review was partial.

The verdict is benign, not a guarantee of safety. Before installing or running its Docker examples, check that the image is pinned to a trusted version or digest and consider that the project directory is mounted writable. The scan was partial, so it did not resolve all referenced paths or assess GIF binary behavior.

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
  • Output HandlingUnvalidated Output Injection, Cross-Context Output, Unbounded Output
  • Tool MisuseTool Parameter Abuse, Chaining Abuse, Unsafe Defaults
  • 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 (22)

Docker image referenced without tag or digest: 'docker run --rm`'.

Medium
Category
MCP Rug Pull
Confidence
75% confidence
Finding

Docker image references without a specific tag (:latest is implicit) or digest (@sha256:...) can be silently replaced by a malicious image.

Content

No source excerpt is available for this finding.

Docker image referenced without tag or digest: 'docker run --rm'.

Medium
Category
MCP Rug Pull
Confidence
82% confidence
Finding

The historical example actually uses an untagged image, which resolves to a mutable latest tag. If copied and executed with an unexpected image, its writable project mount could expose or alter project files; its location in an old changelog substantially reduces the immediate risk.

Content

No source excerpt is available for this finding.

Docker image referenced without tag or digest: 'docker run --rm'.

Medium
Category
MCP Rug Pull
Confidence
75% confidence
Finding

Docker image references without a specific tag (:latest is implicit) or digest (@sha256:...) can be silently replaced by a malicious image.

Content

No source excerpt is available for this finding.

Docker image referenced without tag or digest: 'docker run --rm'.

Medium
Category
MCP Rug Pull
Confidence
75% confidence
Finding

Docker image references without a specific tag (:latest is implicit) or digest (@sha256:...) can be silently replaced by a malicious image.

Content

No source excerpt is available for this finding.

Docker image referenced without tag or digest: 'docker run --rm'.

Medium
Category
MCP Rug Pull
Confidence
75% confidence
Finding

Docker image references without a specific tag (:latest is implicit) or digest (@sha256:...) can be silently replaced by a malicious image.

Content

No source excerpt is available for this finding.

Docker image referenced without tag or digest: 'docker run --rm'.

Medium
Category
MCP Rug Pull
Confidence
75% confidence
Finding

Docker image references without a specific tag (:latest is implicit) or digest (@sha256:...) can be silently replaced by a malicious image.

Content

No source excerpt is available for this finding.

Docker image referenced without tag or digest: 'docker run --rm'.

Medium
Category
MCP Rug Pull
Confidence
75% confidence
Finding

Docker image references without a specific tag (:latest is implicit) or digest (@sha256:...) can be silently replaced by a malicious image.

Content

No source excerpt is available for this finding.

Docker image referenced without tag or digest: 'docker run --rm'.

Medium
Category
MCP Rug Pull
Confidence
75% confidence
Finding

Docker image references without a specific tag (:latest is implicit) or digest (@sha256:...) can be silently replaced by a malicious image.

Content

No source excerpt is available for this finding.

Docker image referenced without tag or digest: 'docker run --rm'.

Medium
Category
MCP Rug Pull
Confidence
75% confidence
Finding

Docker image references without a specific tag (:latest is implicit) or digest (@sha256:...) can be silently replaced by a malicious image.

Content

No source excerpt is available for this finding.

Docker image referenced without tag or digest: 'docker run --rm'.

Medium
Category
MCP Rug Pull
Confidence
75% confidence
Finding

Docker image references without a specific tag (:latest is implicit) or digest (@sha256:...) can be silently replaced by a malicious image.

Content

No source excerpt is available for this finding.

Tool Parameter Abuse

Low
Category
Tool Misuse
Confidence
15% confidence
Finding

Tool parameters are crafted to achieve unintended or unsafe behavior. Parameter abuse can bypass intended safety checks (e.g. shell=True, --force, dangerous glob patterns).

Content

Scanner excerpt · CHANGELOG.md (reported line 132)May include surrounding context.

md
- 输出彩色文本报告或 `--json`;评分 0–100(A/B/C/D);有 FAIL → 退出码 1(可 CI 卡关)。
- 零运行时依赖(仅 Python 3 标准库);只读,绝不读取 `.env` / `*.key` / `*.pem`。
- **`SKILL.md` 定性复核层**:在 Claude Code 内说「检查我的 CLAUDE.md」即可机检 + 模型复核 + 代修复。
- **Docker**:`docker run --rm -v "$PWD:/work" claude-md-doctor`,免装 Python。
- **GitHub Actions CI**:语法 + good/bad 夹具 + JSON 合法性。
- 中英双语 README、MIT LICENSE、CONTRIBUTING、示例动图(CLI 实录 + 动效报告卡)。

Tool Parameter Abuse

Low
Category
Tool Misuse
Confidence
15% confidence
Finding

Tool parameters are crafted to achieve unintended or unsafe behavior. Parameter abuse can bypass intended safety checks (e.g. shell=True, --force, dangerous glob patterns).

Content

Scanner excerpt · Dockerfile (reported line 15)May include surrounding context.

dockerfile
# 把你的项目挂到 /work 即可体检(示例锁定版本 tag,别用无 tag 镜像):
#   docker build -t hekouwang-claude-md-doctor:1.3.7 .
#   docker run --rm -v "$PWD:/work" hekouwang-claude-md-doctor:1.3.7          # 体检当前项目
#   docker run --rm -v "$PWD:/work" hekouwang-claude-md-doctor:1.3.7 /work --json
ENTRYPOINT ["python3", "/app/check.py"]
CMD ["/work"]

Tool Parameter Abuse

Low
Category
Tool Misuse
Confidence
15% confidence
Finding

Tool parameters are crafted to achieve unintended or unsafe behavior. Parameter abuse can bypass intended safety checks (e.g. shell=True, --force, dangerous glob patterns).

Content

Scanner excerpt · Dockerfile (reported line 16)May include surrounding context.

dockerfile
# 把你的项目挂到 /work 即可体检(示例锁定版本 tag,别用无 tag 镜像):
#   docker build -t hekouwang-claude-md-doctor:1.3.7 .
#   docker run --rm -v "$PWD:/work" hekouwang-claude-md-doctor:1.3.7          # 体检当前项目
#   docker run --rm -v "$PWD:/work" hekouwang-claude-md-doctor:1.3.7 /work --json
ENTRYPOINT ["python3", "/app/check.py"]
CMD ["/work"]

Tool Parameter Abuse

Low
Category
Tool Misuse
Confidence
15% confidence
Finding

Tool parameters are crafted to achieve unintended or unsafe behavior. Parameter abuse can bypass intended safety checks (e.g. shell=True, --force, dangerous glob patterns).

Content

Scanner excerpt · README.en.md (reported line 74)May include surrounding context.

Docker (no Python needed)

bash
# Pull the prebuilt image (auto-published to GHCR on each tag; pin a version to keep behavior stable)
docker run --rm -v "$PWD:/work" ghcr.io/huiyonghkw/hekouwang-claude-md-doctor-skill:1.3.7

# Or build locally
docker build -t hekouwang-claude-md-doctor:1.3.7 .

Tool Parameter Abuse

Low
Category
Tool Misuse
Confidence
15% confidence
Finding

Tool parameters are crafted to achieve unintended or unsafe behavior. Parameter abuse can bypass intended safety checks (e.g. shell=True, --force, dangerous glob patterns).

Content

Scanner excerpt · README.en.md (reported line 78)May include surrounding context.

Or build locally

docker build -t hekouwang-claude-md-doctor:1.3.7 . docker run --rm -v "$PWD:/work" hekouwang-claude-md-doctor:1.3.7 # check current project docker run --rm -v "$PWD:/work" hekouwang-claude-md-doctor:1.3.7 /work --json

text

Tool Parameter Abuse

Low
Category
Tool Misuse
Confidence
15% confidence
Finding

Tool parameters are crafted to achieve unintended or unsafe behavior. Parameter abuse can bypass intended safety checks (e.g. shell=True, --force, dangerous glob patterns).

Content

Scanner excerpt · README.en.md (reported line 79)May include surrounding context.

Or build locally

docker build -t hekouwang-claude-md-doctor:1.3.7 . docker run --rm -v "$PWD:/work" hekouwang-claude-md-doctor:1.3.7 # check current project docker run --rm -v "$PWD:/work" hekouwang-claude-md-doctor:1.3.7 /work --json

text

### Gate your PRs (GitHub Actions)

Tool Parameter Abuse

Low
Category
Tool Misuse
Confidence
15% confidence
Finding

Tool parameters are crafted to achieve unintended or unsafe behavior. Parameter abuse can bypass intended safety checks (e.g. shell=True, --force, dangerous glob patterns).

Content

Scanner excerpt · README.md (reported line 75)May include surrounding context.

Docker(不想装 Python 也能跑)

bash
# 拉官方镜像直接用(打 tag 时 GitHub Actions 自动发布到 GHCR,用具体版本号锁住行为)
docker run --rm -v "$PWD:/work" ghcr.io/huiyonghkw/hekouwang-claude-md-doctor-skill:1.3.7

# 或本地自建
docker build -t hekouwang-claude-md-doctor:1.3.7 .

Tool Parameter Abuse

Low
Category
Tool Misuse
Confidence
15% confidence
Finding

Tool parameters are crafted to achieve unintended or unsafe behavior. Parameter abuse can bypass intended safety checks (e.g. shell=True, --force, dangerous glob patterns).

Content

Scanner excerpt · README.md (reported line 79)May include surrounding context.

或本地自建

docker build -t hekouwang-claude-md-doctor:1.3.7 . docker run --rm -v "$PWD:/work" hekouwang-claude-md-doctor:1.3.7 # 体检当前项目 docker run --rm -v "$PWD:/work" hekouwang-claude-md-doctor:1.3.7 /work --json

text

Tool Parameter Abuse

Low
Category
Tool Misuse
Confidence
15% confidence
Finding

Tool parameters are crafted to achieve unintended or unsafe behavior. Parameter abuse can bypass intended safety checks (e.g. shell=True, --force, dangerous glob patterns).

Content

Scanner excerpt · README.md (reported line 80)May include surrounding context.

或本地自建

docker build -t hekouwang-claude-md-doctor:1.3.7 . docker run --rm -v "$PWD:/work" hekouwang-claude-md-doctor:1.3.7 # 体检当前项目 docker run --rm -v "$PWD:/work" hekouwang-claude-md-doctor:1.3.7 /work --json

text

### 接进 CI 卡关(GitHub Actions 示例)

The skill mandates English for code and comments without user opt-in.

Low
Category
Output Handling
Confidence
98% confidence
Finding

Although reporting follows the user's language, the same instruction requires code and comments to use English. This imposes a specific natural language on comments without offering a choice or documenting a justification.

Content

No source excerpt is available for this finding.

The skill requires English code comments without user opt-in.

Low
Category
Output Handling
Confidence
97% confidence
Finding

Although reports follow the user's input language, this line separately requires code and comments to use English. The comment requirement mandates a specific natural language without offering a choice or documenting a justification.

Content

No source excerpt is available for this finding.

Requires English code comments without user opt-in.

Low
Category
Output Handling
Confidence
94% confidence
Finding

Although reports follow the user's language, this line separately requires code and comments to use English. The comment requirement imposes a specific natural language without offering a choice or documenting a justification.

Content

No source excerpt is available for this finding.