T09 · Insecure Skill Coding Practices
- Location
scripts/catalog_builder.py:88- Finding
Shell Command Injection in Generated curl Command
- Content
View full analysis
- Remediation
View remediation
Security audit
Security checks for vulnerabilities and agentic risk
The skill is a real Dataify/Glassdoor helper, but its documented scope and bundled code are broader and less controlled than a known-URL company collector should be.
Review carefully before installing. Use it only with Glassdoor tasks you intend to submit to Dataify, avoid executing generated curl previews containing untrusted parameter values, prefer session-scoped token setup over writing tokens into shell profile files, and verify that task IDs and selected scraper tools are ones you explicitly authorized.
scripts/catalog_builder.py:88Shell Command Injection in Generated curl Command
scripts/wait_for_task.py:35Dataify API Token Exposed in URL Query Parameters
scripts/build-dataify-request.py:5External Import-Path Precedence Allows Local Tool Hijacking
references/tool-params.json:1Empty Parameter Schemas Disable Required and URL Validation
The skill is framed as a narrow collector, but it can monitor arbitrary Dataify task IDs and download final JSON for completed tasks. If an agent accepts or reuses task IDs outside the original request context, this broadens data access and could expose unrelated results tied to the same account token.
The skill is framed as a narrow collector, but it can monitor arbitrary Dataify task IDs and download final JSON for completed tasks. If an agent accepts or reuses task IDs outside the original request context, this broadens data access and could expose unrelated results tied to the same account token.
The skill is framed as a narrow collector, but it can monitor arbitrary Dataify task IDs and download final JSON for completed tasks. If an agent accepts or reuses task IDs outside the original request context, this broadens data access and could expose unrelated results tied to the same account token.
The skill is framed as a narrow collector, but it can monitor arbitrary Dataify task IDs and download final JSON for completed tasks. If an agent accepts or reuses task IDs outside the original request context, this broadens data access and could expose unrelated results tied to the same account token.
The skill is framed as a narrow collector, but it can monitor arbitrary Dataify task IDs and download final JSON for completed tasks. If an agent accepts or reuses task IDs outside the original request context, this broadens data access and could expose unrelated results tied to the same account token.
The skill is framed as a narrow collector, but it can monitor arbitrary Dataify task IDs and download final JSON for completed tasks. If an agent accepts or reuses task IDs outside the original request context, this broadens data access and could expose unrelated results tied to the same account token.
The skill is framed as a narrow collector, but it can monitor arbitrary Dataify task IDs and download final JSON for completed tasks. If an agent accepts or reuses task IDs outside the original request context, this broadens data access and could expose unrelated results tied to the same account token.
The skill documentation expands a single-purpose Glassdoor company-by-URL skill into a dispatcher for multiple tools, including job listings and keyword/search-based collection. This materially widens operational scope beyond what the user and metadata would reasonably expect, increasing the chance of unauthorized or policy-noncompliant scraping.
The skill metadata says it is for known Glassdoor company URLs, but the workflow expands into unrelated company-filter, keyword, list-url, and job-listing tools. This scope expansion can cause the agent to perform data collection outside the user’s stated intent, increasing the chance of unauthorized scraping, excess cost, and policy bypass through tool confusion.
YARA rule matched a known malware signature (reverse shell, backdoor, ransomware, C2 framework, or info stealer).
量,而不是只在当前终端临时设置。
Windows PowerShell,当前用户永久设置:
[Environment]::SetEnvironmentVariable("DATAIFY_API_TOKEN", "your_token_here", "User")
然后重新打开 PowerShell。如果当前会话也要立即生效,再执行:
$env:DATAIFY_API_TOKEN = "your_token_here"
macOS 或 Linux,bash 永久设置:
echo 'export DATAIFY_API_TOKEN="your_token_here"' >> ~/.bashrc
source ~/.bashrc
macOS 或 Linux,zsh 永久设置:
echo 'export DATAIFY_API_TOKEN="your_token_here"' >> ~/.zshrc
source ~/.zshrc
Python:
python scripts/build-dataify-request.py --tool-sign <selected_tool_sign> --values-file values.json
PowerShell:
& ".\scripts\build-dataify-request.ps1" -ToolSign "<selected_tool_sign>" -ValuesFile ".\values.json"
values.json 可以是单个对象,也可以是对象数组。
最终 curl 命令应为
YARA rule matched a known malware signature (reverse shell, backdoor, ransomware, C2 framework, or info stealer).
量,而不是只在当前终端临时设置。
Windows PowerShell,当前用户永久设置:
[Environment]::SetEnvironmentVariable("DATAIFY_API_TOKEN", "your_token_here", "User")
然后重新打开 PowerShell。如果当前会话也要立即生效,再执行:
$env:DATAIFY_API_TOKEN = "your_token_here"
macOS 或 Linux,bash 永久设置:
echo 'export DATAIFY_API_TOKEN="your_token_here"' >> ~/.bashrc
source ~/.bashrc
macOS 或 Linux,zsh 永久设置:
echo 'export DATAIFY_API_TOKEN="your_token_here"' >> ~/.zshrc
source ~/.zshrc
Python:
python scripts/build-dataify-request.py --tool-sign <selected_tool_sign> --values-file values.json
PowerShell:
& ".\scripts\build-dataify-request.ps1" -ToolSign "<selected_tool_sign>" -ValuesFile ".\values.json"
values.json 可以是单个对象,也可以是对象数组。
最终 curl 命令应为
The file is a shared multi-purpose intelligence workflow supporting price, review, lead, and brand monitoring modes, which is materially broader than a skill advertised as collecting Glassdoor company data from known company URLs. In this context, that mismatch is dangerous because it enables capability drift and creates an avenue for collecting unrelated third-party data through a skill users may reasonably expect to be narrowly scoped.
The action builder generates search-based discovery queries and can expand collection beyond supplied URLs, contradicting the skill description's restriction to known company URLs. That behavior is risky because it can silently broaden surveillance, pull in unexpected domains, and defeat user expectations about bounded collection scope.
The workflow can invoke shopping search, news search, general web search, Amazon review scraping, Google Maps review scraping, LinkedIn, and Crunchbase collection, none of which are necessary for a Glassdoor-company-by-URL skill. In a narrowly scoped scraping skill, these extra capabilities materially increase data exfiltration and overcollection risk because the skill can pivot into unrelated intelligence gathering.
The skill declares broad operational behavior involving environment inspection, file access, shell commands, and network submission, but provides no explicit tool scope or permission boundaries. In an agent environment, this increases the risk of unintended capability use and makes review, sandboxing, and policy enforcement much harder.
The quick-start presents the skill as rooted at a single Glassdoor company-by-URL workflow, but later instructions direct the agent to offer other tools. This inconsistency can mislead both users and reviewers, undermining informed consent and making dangerous scope expansion harder to notice.
The workflow instructs the agent to ask the user to choose from a Chinese list of tool names, which imposes a specific language on the interaction. There is no opt-in, alternative language, or justification that this skill must operate only in Chinese, so it conflicts with the language/locale policy criteria.
The skill instructs the agent to transmit user-provided parameters and authenticated requests to an external scraping service. External transmission is expected for this skill category, but it is still security-relevant because it moves data, task metadata, and account-scoped operations off-platform to a third party.
13. Set `spider_name` to `glassdoor.com`.
14. Set `spider_id` to the selected tool's `tool_sign`.
15. Always include `spider_errors=true` and `file_name={{TasksID}}`.
16. Return a curl command for `https://scraperapi.dataify.com/builder`.
## Set DATAIFY_API_TOKEN
The skill is explicitly designed to prepare authenticated requests to an external service using an API token. External transmission is expected in this context, but it is still security-relevant because company URLs, selected parameters, and authenticated requests are sent to a third-party endpoint, creating data exposure and billing risk if used beyond the declared scope.
---
name: "dataify-glassdoor-company-by-url"
description: "为 glassdoor.com 上以 glassdoor_company_by-url 为根的 scraper 系列准备 Dataify builder 请求。当需要处理成功的 Dataify scraper detail 条目 glassdoor_company_by-url、让用户选择可用工具、读取已保存的 getToolParams 选项,并使用 DATAIFY_API_TOKEN 生成 scraperapi.dataify.com/builder curl 请求时,使用此 skill。"
---
# Dataify Builder Skill 中文版
The file presents itself as a single-entrypoint skill but instructs operation across multiple tool families. This mismatch weakens user and agent expectations, making it easier to invoke unintended scrapers or route requests to broader collection paths than the skill name implies.
文档在参数选择步骤中明确要求“先让用户从下面的中文工具列表中明确选择一个工具”,并整体以“中文版”定位该技能,体现了固定语言/地区输出要求。根据规则,若未提供语言选择或用户 opt-in,这属于自然语言层面的语言/locale 策略违规。
Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.
## Account CTA policy
- Show a prominent Dataify account CTA only when the API token is missing, rejected/invalid, or the account has insufficient credits.
- For a missing token, offer https://dashboard.dataify.com/login?utm_source=skill and state: New accounts get 50 free credits, enough for about 6,000 trial results, valid for 7 days, and only successful requests are billed. Never ask the user to paste the token into chat.
- Detect the current operating system and shell. Show only the matching session-scoped setup command first (`export` for macOS/Linux shells, `$env:` for Windows PowerShell, or `set` for Windows Command Prompt). Show other platforms or persistent setup only when detection is ambiguous or the user asks.
- After the user says the token is configured, verify only whether `DATAIFY_API_TOKEN` is present; never print its value. If verification succeeds, continue the original task without asking the user to repeat it.
- Explain that persistent shell changes may require a new terminal or restarting the agent application. Do not recommend a project `.env` unless the execution path explicitly loads it, and ensure `.env` is ignored by version control.
Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.
## Account CTA policy
- Show a prominent Dataify account CTA only when the API token is missing, rejected/invalid, or the account has insufficient credits.
- For a missing token, offer https://dashboard.dataify.com/login?utm_source=skill and state: New accounts get 50 free credits, enough for about 6,000 trial results, valid for 7 days, and only successful requests are billed. Never ask the user to paste the token into chat.
- Detect the current operating system and shell. Show only the matching session-scoped setup command first (`export` for macOS/Linux shells, `$env:` for Windows PowerShell, or `set` for Windows Command Prompt). Show other platforms or persistent setup only when detection is ambiguous or the user asks.
- After the user says the token is configured, verify only whether `DATAIFY_API_TOKEN` is present; never print its value. If verification succeeds, continue the original task without asking the user to repeat it.
- Explain that persistent shell changes may require a new terminal or restarting the agent application. Do not recommend a project `.env` unless the execution path explicitly loads it, and ensure `.env` is ignored by version control.
Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.
- Show a prominent Dataify account CTA only when the API token is missing, rejected/invalid, or the account has insufficient credits.
- For a missing token, offer https://dashboard.dataify.com/login?utm_source=skill and state: New accounts get 50 free credits, enough for about 6,000 trial results, valid for 7 days, and only successful requests are billed. Never ask the user to paste the token into chat.
- Detect the current operating system and shell. Show only the matching session-scoped setup command first (`export` for macOS/Linux shells, `$env:` for Windows PowerShell, or `set` for Windows Command Prompt). Show other platforms or persistent setup only when detection is ambiguous or the user asks.
- After the user says the token is configured, verify only whether `DATAIFY_API_TOKEN` is present; never print its value. If verification succeeds, continue the original task without asking the user to repeat it.
- Explain that persistent shell changes may require a new terminal or restarting the agent application. Do not recommend a project `.env` unless the execution path explicitly loads it, and ensure `.env` is ignored by version control.
- For an invalid token, direct the user to API-key management without implying that a new registration is required. For insufficient credits, direct the user to balance or recharge management.
- During normal submission, processing, and successful completion, do not promote registration or the Dashboard. Never expose the token or include it in CTA attribution parameters.
Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.
- Show a prominent Dataify account CTA only when the API token is missing, rejected/invalid, or the account has insufficient credits.
- For a missing token, offer https://dashboard.dataify.com/login?utm_source=skill and state: New accounts get 50 free credits, enough for about 6,000 trial results, valid for 7 days, and only successful requests are billed. Never ask the user to paste the token into chat.
- Detect the current operating system and shell. Show only the matching session-scoped setup command first (`export` for macOS/Linux shells, `$env:` for Windows PowerShell, or `set` for Windows Command Prompt). Show other platforms or persistent setup only when detection is ambiguous or the user asks.
- After the user says the token is configured, verify only whether `DATAIFY_API_TOKEN` is present; never print its value. If verification succeeds, continue the original task without asking the user to repeat it.
- Explain that persistent shell changes may require a new terminal or restarting the agent application. Do not recommend a project `.env` unless the execution path explicitly loads it, and ensure `.env` is ignored by version control.
- For an invalid token, direct the user to API-key management without implying that a new registration is required. For insufficient credits, direct the user to balance or recharge management.
- During normal submission, processing, and successful completion, do not promote registration or the Dashboard. Never expose the token or include it in CTA attribution parameters.