Back to skill

Security audit

SASAC Performance Analyst

Security checks for vulnerabilities and agentic risk

Overview

The skill mostly matches its stated financial benchmarking purpose, but it needs review because it makes unsupported authority claims and handles sensitive financial reports with weak safeguards.

Review the publisher and authority claims before relying on this for financial decisions. Use only data you are authorized to process, prefer redacted or aggregate financial inputs, choose explicit local output paths, and avoid opening or sharing generated HTML reports that contain untrusted user-supplied text until the report escaping issue is fixed.

Vulnerability Patterns
  • 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
  • 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
Findings (1)

T09 · Insecure Skill Coding Practices

Warning
Location
tools/visualization.py:229
Finding

Stored HTML and JavaScript Injection in Generated Performance Reports

Content
View full analysis
绩效评价报告 - {data.get('industry', '')} ``` ```python
{data.get('industry', '')} · {data.get('size', '')}
``` ```python for key, res in ind_results.items(): level_class = f"level-{res['level'][0].lower()}" if res['level'] else "" html += f""" {res.get('indicator', key)} {res.get('value', '-')} {res.get('unit', '')} {res.get('level', '-')} {res.get('score', '-')} """ ``` ```python with open(output_path, 'w', encoding='utf-8') as f: f.write(html) ``` ### Technical Analysis The HTML report generator interpolates report data directly into HTML markup without escaping it for the relevant HTML context. Affected fields include: - `industry` - `size` - Indicator names and keys - Indicator values and units - Evaluation levels and scores - The dynamically constructed `class` attribute These values can originate from the dictionary supplied to `generate_full_report()`. In normal use, the industry value may originate from user-provided analysis parameters and is retained in the diagnosis result. Because the generator does not encode characters such as `<`, `>`, `"`, `'`, and `&`, crafted content can terminate the intended HTML element or attribute and introduce new executable markup. For example, an industry value resembling the following would break out of the report's title element: ```html ``` The resulting payload is persisted in the generated HTML file. It executes when a recipient opens that file in a browser. The generated document has no Content Security Policy that would prevent inline scr ...[truncated 1580 chars]
Remediation
View remediation
``` 7. Add regression tests covering payloads in every dynamic field, including: ```html "> ``` The tests should verify that payloads appear only as encoded text and do not create executable DOM elements. ]]>
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • System Prompt LeakageDirect Leakage, Indirect Extraction, Tool-Based Exfiltration
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Findings (19)

Ae1

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 525)May include surrounding context.

md
- ✅ **交互式查询系统**:新增 `html/sasac_performance_query_2025.html`

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

The markdown advertises comprehensive report generation, automated financial data extraction from IPO prospectuses, and related benchmarking workflows, which can involve processing and exporting sensitive company financial information. There is no accompanying warning about data sensitivity, privacy considerations, or the need to verify authorization before uploading or generating reports.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

This markdown file presents the skill description, setup, and usage solely in Chinese, which effectively forces a specific language for users reading this documentation. The policy allows locale constraints only when they are explicitly justified or when users are offered a language choice; neither is stated in this file.

Content

No source excerpt is available for this finding.

Undeclared Tool Scope

Medium
Category
MCP Least Privilege
Confidence
92% confidence
Finding

The skill documents capabilities that read local files, parse PDFs, and write reports/charts to disk, but it does not declare any explicit tool scope or allowed-tools restrictions. Without least-privilege constraints, an agent/runtime may grant broader filesystem access than users expect, increasing the chance of unintended reads or writes when processing enterprise documents.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
89% confidence
Finding

The skill promotes automatic extraction of financial data from PDFs and report generation, but it does not warn users that uploaded or local prospectuses may contain sensitive enterprise or transaction data and that derived results may be stored on disk. In a finance/IPO context, this omission is more dangerous because the documents are likely to include confidential or market-sensitive information, creating data leakage and retention risks.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The skill manifest describes a SASAC enterprise performance evaluation skill based on the 2025 enterprise performance benchmark tables, focused on benchmarking, diagnosis, improvement suggestions, and report generation. This file instead defines a Hong Kong IPO prospectus financial-indicator database, including listing types and sample IPO companies, which is a materially different dataset and capability domain from SASAC performance evaluation.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

The system prompt is written to operate in Chinese and presents the assistant as a China-specific enterprise performance expert, but it does not explicitly offer users a choice of language or state that the locale restriction is intentional and required. Under the policy, forcing a specific language without user opt-in is a natural-language policy violation.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
89% confidence
Finding

The workflow explicitly solicits enterprise financial and operational data for analysis, but provides no privacy notice, minimization guidance, or warning against submitting confidential, non-public, or regulated information. In a corporate performance evaluation context, users are likely to paste sensitive financials, creating avoidable confidentiality and data-handling risk even though the prompt is not overtly exfiltrating data.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

The module title, description, metric labels, CLI argument help text, and output strings are presented only in Chinese, which effectively imposes a specific language on users. The file does not offer any language selection or explain that the tool is intentionally limited to a China-specific audience as a documented locale policy choice.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
81% confidence
Finding

The top-level description prominently frames the skill in Chinese and around SASAC standards, but does not clarify whether responses, reports, or interactions will be limited to Chinese or whether users can choose another language. This can be a natural-language policy concern when a skill appears to impose a locale or language expectation without explicit opt-in.

Content

No source excerpt is available for this finding.

Missing User Warnings

Low
Category
Not specified by scanner
Confidence
91% confidence
Finding

The examples show charts and full reports being written to local output paths, but the skill never clearly tells users that execution creates files in the output directory. This can lead to unintentional local data persistence, especially when the generated artifacts contain sensitive performance assessments or financial details.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
86% confidence
Finding

This JSON file presents all human-readable titles, descriptions, and case content exclusively in Chinese, indicating a fixed language/locale for the skill data. Under the policy rule, a language constraint should either offer user opt-in/choice or be clearly documented as region-specific and justified; this file contains no such explanation.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
81% confidence
Finding

This JSON file contains all user-facing descriptions, metrics, and recommendations exclusively in Chinese, but it does not document that the skill is China-specific or that language selection is intentional. Under the policy rule for natural-language constraints, this can be a locale/language policy issue because it implicitly forces one language without opt-in or justification.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
91% confidence
Finding

This JSON dataset uses Chinese-only natural-language descriptions and labels, including the top-level description and field values, without any indication that the skill is region-specific or that users can opt into this locale. Under the policy rule for natural-language violations, a file that effectively forces a specific language without user opt-in can be flagged.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
91% confidence
Finding

This JSON file contains a natural-language description in Chinese only, which can impose a specific language/locale on users or downstream tooling without any opt-in or justification. The stated policy requires flagging language or locale constraints unless the file offers a choice or clearly documents a justified region-specific scope.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
91% confidence
Finding

The document declares lang="zh-CN", which fixes the interface to Simplified Chinese/China locale. Under the natural-language policy rules, a forced language/locale can be a policy concern when the file does not offer any user opt-in or explain that the tool is intentionally region-specific.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
78% confidence
Finding

The description is primarily presented in Chinese and English mixed text centered on the PRC SASAC context, but there is no explicit statement that the skill is region-specific or that users can choose their preferred language/locale. Under the policy rule, locale constraints should either be optional or clearly justified as region-specific.

Content

No source excerpt is available for this finding.

Missing User Warnings

Low
Category
Not specified by scanner
Confidence
85% confidence
Finding

The HTML report generator creates a temporary file when no path is provided and writes report contents to disk, but there is no print/log message, confirmation, or explicit warning in the docstring that a file will be created automatically. For a code-file warning check, this is a file-write operation lacking visible disclosure at the point of execution.

Content

No source excerpt is available for this finding.

Missing User Warnings

Low
Category
Not specified by scanner
Confidence
85% confidence
Finding

The text report generator writes data to a file and silently creates a temporary file if no output path is supplied. There is no confirmation prompt, print/log statement, or docstring warning explaining this file write behavior, so users may not realize a local artifact is being created.

Content

No source excerpt is available for this finding.