Back to skill

Security audit

Skill严选猎手

Security checks for vulnerabilities and agentic risk

Overview

The skill mostly matches its search/audit/trending purpose, but it uses risky automatic install paths, broad triggers, persistent scheduling/profile behavior, and weak token handling that users should review first.

Install only after reviewing the shell commands it may run. Avoid curl-to-bash and unpinned npx paths, do not store broad GitHub tokens in ~/.bashrc for this tool, confirm before enabling cron reports, and treat its scanner output as a heuristic rather than a safety guarantee.

Vulnerability Patterns
  • Remote Payload Retrieval and ExecutionFetches external code whose behavior can change after review
  • Insecure DependenciesIntroduces malicious components through unsafe dependency sources
  • 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
Findings (4)

T03 · Remote Payload Retrieval and Execution

Error
Location
SKILL.md:171
Finding

Unverified Remote Installer Is Piped Directly into Bash

Content
View full analysis
Remediation
View remediation
skillhub-install.sh" | sha256sum --check ``` 5. Prefer a signed release and verify the signature against a documented, pinned public key. 6. Display or audit the downloaded script before asking for explicit user approval to execute it. 7. For search-only workflows, use the existing public SkillHub HTTP API and do not install the CLI. 8. Document the files, permissions, and network endpoints used by any installer. ]]>

T08 · Insecure Dependencies

Error
Location
scripts/search.py:103
Finding

Unpinned Packages May Be Downloaded and Executed Automatically Through npx

Content
View full analysis
Remediation
View remediation

T09 · Insecure Skill Coding Practices

Error
Location
scripts/scanner.py:21
Finding

Scanner Domain Allowlist Can Be Bypassed and Does Not Reliably Detect Download-and-Execute Chains

Content
View full analysis
Remediation
View remediation

T09 · Insecure Skill Coding Practices

Warning
Location
scripts/search.py:16
Finding

Search Script Reads a GitHub Token from the User's Shell Startup File

Content
View full analysis
Remediation
View remediation
Vulnerability Patterns
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • Tool MisuseTool Parameter Abuse, Chaining Abuse, Unsafe Defaults
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
Findings (79)

Tp4

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

The skill description emphasizes on-demand search, but the file also defines scheduled trend collection, profile-driven recommendation logic, and persistent local user profiling. That gap hides automation and data persistence behaviors that materially change risk and user expectations.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
95% confidence
Finding

The skill description emphasizes on-demand search, but the file also defines scheduled trend collection, profile-driven recommendation logic, and persistent local user profiling. That gap hides automation and data persistence behaviors that materially change risk and user expectations.

Content

No source excerpt is available for this finding.

Vague Triggers

High
Category
Not specified by scanner
Confidence
97% confidence
Finding

The trigger description is overly broad and includes common phrases like asking for existing tools or whether something exists on GitHub, making accidental invocation likely. Because the skill can run shell/network actions and guide installation flows, unintended activation meaningfully increases attack surface and the chance of unsafe side effects.

Content

No source excerpt is available for this finding.

External Script Fetching

High
Category
Supply Chain
Confidence
97% confidence
Finding

The skill recommends fetching and piping a remote script directly into bash, which allows arbitrary code execution from a remote server without integrity verification. If the hosting endpoint or network path is compromised, the agent or user could execute attacker-controlled code on the host.

Content

Scanner excerpt · SKILL.md (reported line 57)May include surrounding context.

md
**降级处理**:
- GitHub 速率限制 → 提示配置 GITHUB_TOKEN,继续用 ClawHub / SkillHub 结果
- CLI 未安装 → 报告具体安装命令,继续其他源
- SkillHub CLI 未安装 → 提示安装命令(`curl -fsSL https://skillhub-1388575217.cos.ap-guangzhou.myqcloud.com/install/install.sh | bash -s -- --cli-only`),继续其他源

**SkillHub 优先策略**:当 SkillHub CLI 可用时,搜索结果中 SkillHub 来源的 Skill 安装更便捷(国内加速、合规),优先推荐给国内用户。安装命令:`skillhub install <slug> --dir <skills目录>`

Tool Parameter Abuse

High
Category
Tool Misuse
Confidence
90% confidence
Finding

The cleanup command interpolates <skill_name> into rm -rf on a filesystem path. If skill_name is attacker-controlled or insufficiently sanitized, path traversal or argument manipulation could delete unintended files or directories on the host.

Content

Scanner excerpt · SKILL.md (reported line 95)May include surrounding context.

md
| 🔴 HIGH | ❌ 建议拒绝 | 列出 Hard Flags,告知木老师,等待明确指令 |
| ⛔ EXTREME | ❌ 拒绝 | 直接拒绝,清理暂存区,报告木老师 |

4. 扫描完成后清理暂存区:`rm -rf /tmp/skill-hunter-staging/<skill_name>`

**安全约束**:
- 扫描器只输出摘要(rule_id / rule_name / 文件名 / 行号),不输出原始代码片段

Tool Parameter Abuse

High
Category
Tool Misuse
Confidence
86% confidence
Finding

The cleanup command interpolates <skill_name> into rm -rf on a filesystem path. If skill_name is attacker-controlled or insufficiently sanitized, path traversal or argument manipulation could delete unintended files or directories on the host.

Content

Scanner excerpt · SKILL.md (reported line 95)May include surrounding context.

md
| 🔴 HIGH | ❌ 建议拒绝 | 列出 Hard Flags,告知木老师,等待明确指令 |
| ⛔ EXTREME | ❌ 拒绝 | 直接拒绝,清理暂存区,报告木老师 |

4. 扫描完成后清理暂存区:`rm -rf /tmp/skill-hunter-staging/<skill_name>`

**安全约束**:
- 扫描器只输出摘要(rule_id / rule_name / 文件名 / 行号),不输出原始代码片段

External Script Fetching

High
Category
Supply Chain
Confidence
97% confidence
Finding

The automatic setup flow includes curl ... | bash for SkillHub installation, making first-use bootstrap capable of arbitrary remote code execution. This is more dangerous in context because it is embedded in an automated dependency path that may run with little scrutiny.

Content

Scanner excerpt · SKILL.md (reported line 151)May include surrounding context.

md
npx clawhub --version 2>/dev/null && echo "✅ clawhub CLI" || (echo "⚠️ 正在安装 clawhub..." && npm i -g clawhub && echo "✅ clawhub 安装完成")

# 源3: SkillHub(腾讯,独立 CLI,无需鉴权)
command -v skillhub && skillhub --version 2>/dev/null && echo "✅ skillhub CLI" || (echo "⚠️ 正在安装 skillhub..." && curl -fsSL https://skillhub-1388575217.cos.ap-guangzhou.myqcloud.com/install/install.sh | bash -s -- --cli-only && echo "✅ skillhub 安装完成")

# 源4: Skills.sh(需 Node.js + npm,无需鉴权)
npx skills --version 2>/dev/null && echo "✅ skills CLI" || (echo "⚠️ 正在安装 skills..." && npm i -g skills && echo "✅ skills 安装完成")

Chaining Abuse

High
Category
Tool Misuse
Confidence
90% confidence
Finding

The chained bootstrap command mixes checks, fallbacks, remote fetching, and shell execution in a single expression. This makes failure handling opaque and increases the chance that a partial check or unexpected environment condition results in unintended execution of a remote installer.

Content

Scanner excerpt · SKILL.md (reported line 151)May include surrounding context.

md
npx clawhub --version 2>/dev/null && echo "✅ clawhub CLI" || (echo "⚠️ 正在安装 clawhub..." && npm i -g clawhub && echo "✅ clawhub 安装完成")

# 源3: SkillHub(腾讯,独立 CLI,无需鉴权)
command -v skillhub && skillhub --version 2>/dev/null && echo "✅ skillhub CLI" || (echo "⚠️ 正在安装 skillhub..." && curl -fsSL https://skillhub-1388575217.cos.ap-guangzhou.myqcloud.com/install/install.sh | bash -s -- --cli-only && echo "✅ skillhub 安装完成")

# 源4: Skills.sh(需 Node.js + npm,无需鉴权)
npx skills --version 2>/dev/null && echo "✅ skills CLI" || (echo "⚠️ 正在安装 skills..." && npm i -g skills && echo "✅ skills 安装完成")

External Script Fetching

High
Category
Supply Chain
Confidence
96% confidence
Finding

The installation guide instructs users to execute a remotely fetched shell script directly. This normalizes an unsafe installation practice and creates a straightforward compromise path if the remote asset changes or is intercepted.

Content

Scanner excerpt · SKILL.md (reported line 171)May include surrounding context.

md
- 官网:https://clawhub.ai

**源3 · SkillHub**(腾讯,独立 CLI,无需鉴权)
- 安装:`curl -fsSL https://skillhub-1388575217.cos.ap-guangzhou.myqcloud.com/install/install.sh | bash -s -- --cli-only`
- 验证:`skillhub --version`
- 用途:`skillhub search <q>` 搜索、`skillhub install <slug> --dir <目录>` 安装
- search.py / trending.py 也支持直接调用 SkillHub 公开 HTTP API(`https://api.skillhub.cn/api/skills`),无需安装 CLI

Ae1

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 193)May include surrounding context.

md
必读文件:`./SKILL.md`

External Script Fetching

High
Category
Supply Chain
Confidence
99% confidence
Finding

This is a direct instance of external script fetching and execution via curl-to-bash. In the context of a skill whose purpose is discovering and installing external tools, this is more dangerous because it normalizes trust in third-party code and could lead users to execute attacker-controlled installers or supply-chain-compromised content.

Content

Scanner excerpt · references/search-guide.md (reported line 54)May include surrounding context.

CLI 安装:

bash
curl -fsSL https://skillhub-1388575217.cos.ap-guangzhou.myqcloud.com/install/install.sh | bash -s -- --cli-only

搜索命令:

Credential Access

High
Category
Privilege Escalation
Confidence
95% confidence
Finding

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

Content

Scanner excerpt · references/security-levels.md (reported line 21)May include surrounding context.

md
| R1 | 外部 URL 请求 | curl/wget 到非白名单域名 | 数据外泄 |
| R2 | base64 解码执行 | base64 -d / atob() | 隐藏 payload |
| R3 | eval/exec 动态执行 | eval() / exec() | 远程代码执行 |
| R4 | 读取系统凭证路径 | ~/.ssh / ~/.aws / /etc/shadow | 凭证窃取 |
| R5 | 访问 Agent 私密文件 | MEMORY.md / SOUL.md / agent-config.json | 隐私泄露 |
| R6 | 混淆/超长单行代码 | 单行 >500 字符 | 隐藏恶意逻辑 |
| R7 | 裸 IP 网络请求 | http://x.x.x.x | 绕过域名审查 |

Credential Access

High
Category
Privilege Escalation
Confidence
95% confidence
Finding

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

Content

Scanner excerpt · scripts/scanner.py (reported line 44)May include surrounding context.

python
| R1 | 外部 URL 请求 | curl/wget 到非白名单域名 | 数据外泄 |
| R2 | base64 解码执行 | base64 -d / atob() | 隐藏 payload |
| R3 | eval/exec 动态执行 | eval() / exec() | 远程代码执行 |
| R4 | 读取系统凭证路径 | ~/.ssh / ~/.aws / /etc/shadow | 凭证窃取 |
| R5 | 访问 Agent 私密文件 | MEMORY.md / SOUL.md / agent-config.json | 隐私泄露 |
| R6 | 混淆/超长单行代码 | 单行 >500 字符 | 隐藏恶意逻辑 |
| R7 | 裸 IP 网络请求 | http://x.x.x.x | 绕过域名审查 |

Tool Parameter Abuse

High
Category
Tool Misuse
Confidence
80% 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 · references/security-levels.md (reported line 25)May include surrounding context.

md
| R5 | 访问 Agent 私密文件 | MEMORY.md / SOUL.md / agent-config.json | 隐私泄露 |
| R6 | 混淆/超长单行代码 | 单行 >500 字符 | 隐藏恶意逻辑 |
| R7 | 裸 IP 网络请求 | http://x.x.x.x | 绕过域名审查 |
| R8 | 提权操作 | sudo / chmod 777 / chown root | 提权攻击 |
| R9 | 凭证外发 | token/key/secret + curl/fetch | 凭证泄露 |
| R10 | 未声明包安装 | pip/npm/apt install 未在文档说明 | 供应链投毒 |
| AI1 | Prompt Injection | "ignore previous instructions" 等 | 隐藏指令攻击 |

Chaining Abuse

High
Category
Tool Misuse
Confidence
75% confidence
Finding

Tool calls are chained to bypass individual safety checks or escalate capabilities beyond what any single tool call would allow.

Content

Scanner excerpt · references/security-levels.md (reported line 25)May include surrounding context.

md
| R5 | 访问 Agent 私密文件 | MEMORY.md / SOUL.md / agent-config.json | 隐私泄露 |
| R6 | 混淆/超长单行代码 | 单行 >500 字符 | 隐藏恶意逻辑 |
| R7 | 裸 IP 网络请求 | http://x.x.x.x | 绕过域名审查 |
| R8 | 提权操作 | sudo / chmod 777 / chown root | 提权攻击 |
| R9 | 凭证外发 | token/key/secret + curl/fetch | 凭证泄露 |
| R10 | 未声明包安装 | pip/npm/apt install 未在文档说明 | 供应链投毒 |
| AI1 | Prompt Injection | "ignore previous instructions" 等 | 隐藏指令攻击 |

Instruction Override

High
Category
Prompt Injection
Confidence
80% confidence
Finding

This pattern attempts to override system instructions or ignore safety constraints. Without LLM analysis, manual review is recommended.

Content

Scanner excerpt · references/security-levels.md (reported line 28)May include surrounding context.

md
| R8 | 提权操作 | sudo / chmod 777 / chown root | 提权攻击 |
| R9 | 凭证外发 | token/key/secret + curl/fetch | 凭证泄露 |
| R10 | 未声明包安装 | pip/npm/apt install 未在文档说明 | 供应链投毒 |
| AI1 | Prompt Injection | "ignore previous instructions" 等 | 隐藏指令攻击 |
| AI2 | 挖矿特征 | xmrig / stratum+tcp / coinhive | 恶意占用算力 |

## Yellow Flag 规则(Y1~Y4)

Description-Behavior Mismatch

High
Category
Not specified by scanner
Confidence
98% confidence
Finding

The file implements a local static security scanner, which materially differs from the skill’s declared purpose of external skill discovery/search. In an agent ecosystem, undeclared security-review behavior can change what data is read, processed, and reported, creating trust and scope-boundary issues even if the code is not overtly malicious.

Content

No source excerpt is available for this finding.

Credential Access

High
Category
Privilege Escalation
Confidence
60% confidence
Finding

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

Content

Scanner excerpt · scripts/scanner.py (reported line 44)May include surrounding context.

python
{
        "id": "R4",
        "name": "读取系统凭证路径",
        "pattern": r'(~/\.ssh|~/\.aws|~/\.config/gcloud|~/\.gnupg|/etc/shadow|/etc/passwd)',
        "desc": "访问系统凭证目录(凭证窃取风险)",
        "level": "🚨 HARD REJECT",
    },

YARA rule 'crypto_miner_software': References to known cryptocurrency mining software [cryptominers]

High
Category
YARA Match
Confidence
80% confidence
Finding

YARA rule matched cryptocurrency mining indicators (stratum protocol, mining pools, miner binaries, or cryptojacking scripts).

Content

Scanner excerpt · scripts/scanner.py (reported line 104)May include surrounding context.

python
"pattern": r'(ignore\s+(previous|all|prior)\s+instructions?|forget\s+(everything|all)|you\s+are\s+now\s+a|disregard\s+(your|all)|new\s+instructions?:)',
        "desc": "Prompt Injection 攻击(隐藏指令)",
        "level": "🚨 HARD REJECT",
        "case_insensitive": True,
    },
    {
        "id": "AI2",
        "name": "挖矿特征",
        "pattern": r'(stratum\+tcp|mining\.pool|xmrig|cryptonight|monero|coinhive)',
        "desc": "加密货币挖矿特征",
        "level": "🚨 HARD REJECT",
        "case_insensitive": True,
    },
]

YELLOW_FLAG_RULES = [
    {
        "id": "Y1",
        "name": "写入 Agent 持久化目录",
        "pattern": r'~/\.[a-z]+/',
        "desc": "写入 Agent 持久化目录(需确认写入范围是否合理)",
        "level": "⚠️ 需人工确认",
    },
    {
        "id": "Y2",
        "name": "已知外部网络请求",
        "pattern": r'(requests\.get|urllib\.request|fetch\(|axios\.get|http\.get)',
        "desc":

External Script Fetching

High
Category
Supply Chain
Confidence
91% confidence
Finding

The script text recommends installing a CLI by piping a remote script directly into bash, an unsafe pattern that enables arbitrary code execution if the hosting endpoint or transport chain is compromised. Even though this line is documentation rather than an executed statement, embedding such guidance in a skill encourages insecure operator behavior.

Content

Scanner excerpt · scripts/search.py (reported line 197)May include surrounding context.

python
优先使用公开 HTTP API(无需鉴权),CLI 作为回退
    API: GET https://api.skillhub.cn/api/skills?keyword=<query>&sortBy=score&pageSize=<N>
    CLI: skillhub search <query> --json
    CLI 安装: curl -fsSL https://skillhub-1388575217.cos.ap-guangzhou.myqcloud.com/install/install.sh | bash -s -- --cli-only
    """
    # 优先使用 HTTP API(公开,无需鉴权)
    try:

External Script Fetching

High
Category
Supply Chain
Confidence
91% confidence
Finding

This error message again instructs users to run a remote script through bash, reinforcing an insecure installation path. Because the skill's purpose is to help users discover and use external tools, unsafe install guidance materially increases the likelihood of host compromise.

Content

Scanner excerpt · scripts/search.py (reported line 240)May include surrounding context.

python
# CLI 回退
    bin_path = _find_skillhub_bin()
    if not bin_path:
        return [], f"SkillHub API 不可用({api_err_msg}),CLI 未安装(安装: curl -fsSL https://skillhub-1388575217.cos.ap-guangzhou.myqcloud.com/install/install.sh | bash -s -- --cli-only)"
    try:
        result = subprocess.run(
            [bin_path, "search", query, "--json", "--search-limit", str(min(limit, 20))],

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The trigger logic is explicitly broad enough to activate not only on direct requests to search external skills, but also on vague phrases like asking for external tools or existing skills. Because this skill searches and stages third-party code from multiple external sources, over-broad activation increases the chance it is invoked in contexts where the user did not clearly consent to external discovery or supply-chain exposure. In this context, the breadth is more dangerous than usual because the skill can lead users toward downloading and installing untrusted external skills.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

The trigger conditions are intentionally broad enough to activate on generic phrases like '找外部工具' or '有没有现成的Skill', which can cause the skill to run in contexts the user did not explicitly intend. In a security-sensitive environment, overbroad invocation can lead to unnecessary external lookups, increased exposure of user intent to third-party services, and accidental workflow hijacking away from safer internal tooling.

Content

No source excerpt is available for this finding.

Intent-Code Divergence

Medium
Category
Not specified by scanner
Confidence
99% confidence
Finding

The README claims the tool 'does not send data to any server', but elsewhere describes querying GitHub, SkillHub, and other external platforms. This kind of false privacy/safety representation can mislead users into supplying sensitive search terms, repository paths, or tokens under the assumption that no network activity occurs, which is especially risky for a tool centered on discovering external skills.

Content

No source excerpt is available for this finding.

Undeclared Tool Scope

Medium
Category
MCP Least Privilege
Confidence
95% confidence
Finding

The skill advertises and documents use of environment variables, local file access, network access, and shell execution, but it does not declare any explicit tool scope such as permissions or allowed-tools. That creates a least-privilege failure: an invocation of this skill could access broader capabilities than reviewers or users can easily verify from the manifest.

Content

No source excerpt is available for this finding.