Back to skill

Security audit

My

Security checks for vulnerabilities and agentic risk

Overview

The skill’s news-report purpose is plausible, but its included generator has unsafe local file and link handling plus hard-coded output paths that should be reviewed before installation.

Review this skill before installing or running it. It is meant to fetch public Chinese tech news and write a local HTML report, but the bundled script should be patched to use user-configurable paths, validate image keys, allow only http/https article links, and avoid using untrusted JSON input without checks.

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 (2)

T09 · Insecure Skill Coding Practices

Warning
Location
scripts/gen_html.py:29
Finding

Path Traversal Allows Unauthorized Local Image Disclosure

Content
View full analysis
Remediation
View remediation

T09 · Insecure Skill Coding Practices

Warning
Location
scripts/gen_html.py:114
Finding

Unvalidated URL Schemes Permit Unsafe Links in Generated Reports

Content
View full analysis
{img_tag}

{esc(title_sp)}

{esc(text_sp)}

📡 {esc(src)} ``` ### Technical Analysis The `esc()` function HTML-escapes special characters, which prevents direct attribute termination and markup injection. It does not, however, establish whether the supplied URL uses a safe protocol. Because `url` comes from the input JSON, an attacker can provide values using schemes such as `javascript:` or `data:` rather than the expected HTTP or HTTPS article URL. The resulting value remains syntactically intact after HTML escaping and is placed directly in the anchor's `href` attribute. Exploitation requires a user to click the generated source link. Exact handling of non-HTTP schemes depends on the browser and its security policy, but allowing arbitrary schemes unnecessarily exposes report readers to script execution, attacker-controlled content, or unintended protocol handlers. The link also uses `target="_blank"` without explicitly specifying `rel="noopener noreferrer"`. Modern browsers commonly provide implicit opener isolation, but an explicit policy is safer and more portable. ### Attack Path 1. An attacker supplies or modifies an input article whose `url` field contains an unsafe scheme or attacker-controlled destination. 2. The generator HTML-escapes the string but performs no scheme or host validation. 3. The unsafe value is written into the generated report's `href` attribute. 4. A recipient opens the report and clicks the source link. 5. The browser processes the supplied scheme or navigates to attacker-controlled content, subjec ...[truncated 777 chars]
Remediation
View remediation
{esc(src)}' ) else: source_markup = f'📡 {esc(src)}' ``` ]]>
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (6)

Missing User Warnings

Low
Category
Not specified by scanner
Confidence
92% confidence
Finding

The skill fetches content from third-party sources (IT之家 RSS, 快科技 pages, and article pages) without warning the user about outbound network access. This is dangerous from a transparency and privacy perspective because execution may disclose network metadata, trigger unexpected external requests, or violate restricted-environment policies even if the underlying purpose is legitimate news aggregation.

Content

No source excerpt is available for this finding.

Missing User Warnings

Low
Category
Not specified by scanner
Confidence
89% confidence
Finding

The skill explicitly instructs downloading images and generating HTML files on the local filesystem, but it does not warn the user that running it will create directories and write output files. While this appears to be normal functionality for a report-generation skill rather than overtly malicious behavior, undisclosed filesystem modification can surprise users, overwrite existing artifacts, or violate execution-environment expectations.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
87% confidence
Finding

The entire document is written only in Chinese and specifies Chinese/CJK-specific formatting rules, with no indication that users may opt into another language or that the skill is explicitly limited to a Chinese-language context. This can violate language/locale policy when a skill implicitly requires a specific language without user choice.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
88% confidence
Finding

The script's docstring, HTML lang attribute, title, and visible page text are all fixed to Chinese-language output such as "科技资讯日报" and "zh-CN". The file does not indicate that this locale is optional, user-selectable, or required for a region-specific purpose.

Content

No source excerpt is available for this finding.

Intent-Code Divergence

Low
Category
Not specified by scanner
Confidence
98% confidence
Finding

The module docstring states the generated HTML will be output 'in the same directory' as the script or data context, but the code later replaces that computed path with a hard-coded workspace path. This is an active contradiction between documented behavior and actual file-write location, not merely omitted detail.

Content

No source excerpt is available for this finding.

Missing User Warnings

Low
Category
Not specified by scanner
Confidence
83% confidence
Finding

This code overrides the computed output location with a hard-coded workspace path and then writes the generated HTML file there. Although the module docstring mentions an output file, the actual forced destination differs from the earlier description and there is no confirmation prompt or inline warning near the write operation.

Content

No source excerpt is available for this finding.