Skill Trust Decision

agent-boarding-pass

代码行为干净(纯文件读写+哈希验签),但 SKILL.md 声明了 Bash/WebFetch 权限而代码未使用,存在文档欺骗(doc_deception)瑕疵,定级为低风险。

Install decision first Source: ClawHub Scanned: 13 days ago
Files 6
Artifacts 2
Violations 2
Findings 3

Why this conclusion was reached

1/4 dimensions flagged
Block
Declared vs actual capability

2 undeclared or violating capabilities were inferred.

Review
Hidden execution and egress

2 lower-risk artifacts were extracted and still need context.

Pass
Attack chain and severe findings

There is no explicit malicious chain in the report.

Review
Dependencies and supply chain hygiene

Dependency information is incomplete, so supply-chain confidence stays limited.

What drove the risk score up

文档声明超出实际能力(doc_deception 轻微) +20

SKILL.md allowed-tools 声明了 Bash/Write/WebFetch,但代码仅用 pathlib+hashlib+json 标准库,零第三方依赖,无任何 shell 调用和网络请求

Write 权限声明宽泛但代码可控 +5

仅输出用户指定路径的 JSON 文件,out 路径由用户控制,无越权写操作

Most important evidence

Medium Doc Mismatch

声明了 Bash 权限但代码未使用

SKILL.md allowed-tools 包含 Bash,但 agent_boarding_pass.py 代码中无任何 subprocess、os.system、os.popen 等 shell 调用。声明超出实际需要,存在权限声明宽泛问题。

SKILL.md:7
从 allowed-tools 中移除 Bash,或在 SKILL.md 中说明 Bash 的具体使用场景(如有未来扩展计划)
Medium Doc Mismatch

声明了 WebFetch 权限但代码未使用

SKILL.md allowed-tools 包含 WebFetch,但代码完全无任何网络请求(无 requests/urllib/socket 等)。

SKILL.md:7
从 allowed-tools 中移除 WebFetch
Low Supply Chain

零依赖实现,无版本锁定风险

代码使用纯 Python 标准库(argparse, datetime, hashlib, json, pathlib, sys),无第三方依赖,无 pip install,可直接运行。

scripts/agent_boarding_pass.py:1
当前状态良好,保持零依赖

Declared capability vs actual capability

Filesystem Pass
Declared READ
→
Inferred READ
scripts/agent_boarding_pass.py:31-32 读取证据文件内容
Filesystem Pass
Declared WRITE
→
Inferred WRITE
scripts/agent_boarding_pass.py:54-55 写登机牌 JSON 到用户指定路径
Shell Block
Declared WRITE
→
Inferred NONE
SKILL.md allowed-tools 声明 Bash,但代码无任何 subprocess/os.system 调用
Network Block
Declared READ
→
Inferred NONE
SKILL.md allowed-tools 声明 WebFetch,但代码无任何网络请求
Browser Pass
Declared NONE
→
Inferred NONE
N/A
Clipboard Pass
Declared NONE
→
Inferred NONE
N/A
Database Pass
Declared NONE
→
Inferred NONE
N/A
Environment Pass
Declared NONE
→
Inferred NONE
N/A
Skill Invoke Pass
Declared NONE
→
Inferred NONE
N/A

Suspicious artifacts and egress

Medium External URL
https://zhaoxinghua09-cell.github.io/lgd-hub/

README.md:20

Medium External URL
https://medxpert.cn/badge/powered/svg/lgd-powered-cn.svg

README.md:38

Dependencies and supply chain

There are no structured dependency warnings.

File composition

6 files · 443 lines
Markdown 4 files · 289 linesPython 1 files · 141 linesJSON 1 files · 13 lines
Files of concern · 3
scripts/agent_boarding_pass.py Python · 141 lines
零依赖实现,无版本锁定风险
README.md Markdown · 123 lines
https://zhaoxinghua09-cell.github.io/lgd-hub/ · https://medxpert.cn/badge/powered/svg/lgd-powered-cn.svg
SKILL.md Markdown · 123 lines
声明了 Bash 权限但代码未使用 · 声明了 WebFetch 权限但代码未使用
Other files · LICENSE.md · ATTESTATION.md · manifest.json

Security positives

代码逻辑透明清晰:SHA-256 防篡改签名 + 过期校验,符合安全设计原则
零第三方依赖,只用 Python 标准库,无供应链风险
无网络请求、无 shell 执行、无凭证读取、无数据外传
错误处理规范(rc=0/1/2 明确,无原始栈帧泄漏)
文件操作有路径安全检查(strip @前缀)
ATTESTATION.md 提供了权属与原创性声明