Back to skill

Security audit

Travel planner case

Security checks for vulnerabilities and agentic risk

Overview

This skill is a Chinese-language travel-planning helper that searches travel platforms and prepares presentation content, with no evidence of hidden persistence, credential access, destructive behavior, or exfiltration.

Install only if you are comfortable with a Chinese-language travel workflow that may send destination searches to public travel/social platforms and may create local presentation-related output. Treat the bundled scripts as scaffolding or JSON generators rather than a complete automated PPT system.

Vulnerability Patterns
  • 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
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • 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
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (18)

Tp4

High
Category
MCP Tool Poisoning
Confidence
94% confidence
Finding

The declared description emphasizes an end-to-end intelligent travel research pipeline: cross-platform discovery, recommendation ranking, activity prioritization, trap avoidance, image collection, and polished PPT generation. The supplied code chunk does not implement those core capabilities. Instead, it is a formatter that turns already-available attraction data into a slide-content JSON structure, optionally loading that data from a local JSON file and otherwise using sample placeholders. While the produced slide structure is related to travel-guide PPT creation, the primary behavior shown is substantially narrower than the declared functionality, so this is a material description-behavior mismatch.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
95% confidence
Finding

The declared description overstates the implemented behavior. While the theme matches travel-guide planning, the code is largely a scaffold/prototype: it accepts a city, creates predefined search keywords, uses mock/hardcoded attraction data, computes simple scores, and formats output for a possible PPT. The primary purpose is adjacent, but key promised capabilities—real multi-platform search, extraction of hot recommendations, detailed activity prioritization based on fetched content, exclusion logic based on actual crowd/坑 feedback, and automatic image-backed PPT creation—are not actually implemented in this chunk. Therefore the description does not accurately represent the code's real behavior.

Content

No source excerpt is available for this finding.

Undeclared Tool Scope

Medium
Category
MCP Least Privilege
Confidence
70% confidence
Finding

Without declared permissions the skill's intent is opaque and cannot be validated.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
90% confidence
Finding

The skill directs external searches across multiple third-party platforms and image retrieval without informing users that network requests will be made. Undisclosed outbound requests can leak user intent, destination queries, and usage metadata to external services, which is more sensitive here because travel plans can reveal personal interests or future movement patterns.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

The skill promises automatic PPT generation and returns a file path without warning that local file creation/write operations may occur. Unexpected file writes can violate user expectations, create clutter or overwrite risks, and become more serious if the runtime allows broad filesystem access or attacker-influenced output paths.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
97% confidence
Finding

This markdown file appears to force a specific language/locale for the skill content, which matches the policy-violation category for language constraints without user opt-in. There is no indication anywhere in the file that Chinese is optional, selectable, or required for a documented region-specific purpose.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

This code file contains user-facing natural-language descriptions entirely in Chinese, including the module docstring and later CLI/help/output strings, but does not indicate that the skill is intended only for Chinese-speaking users or provide any opt-in language selection. Under the policy, forcing a specific language without user choice is a natural-language policy violation.

Content

No source excerpt is available for this finding.

Intent-Code Divergence

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The function documentation states that it searches all platforms and returns platform search results. However, the implementation explicitly notes that results are simulated and only returns structured queries plus empty summary fields, which contradicts the stated behavior rather than merely omitting detail.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

The manifest describes real automated collection of hot recommendations from Xiaohongshu, Weibo, Bilibili, and Douyin, ranking attractions based on those results, and ultimately generating a PPT with network images. In this code, the 'search' is explicitly simulated, attractions are hard-coded examples, and 'generate_ppt_data' only builds a data structure rather than creating a PPT, so the implemented behavior is materially narrower than the stated capability.

Content

No source excerpt is available for this finding.

Intent-Code Divergence

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

The documentation claims the function processes raw search result data and extracts attraction information. In reality, it does not read from 'raw_data' at all and instead scores a fixed in-code list, which is a direct contradiction of the declared intent.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
97% confidence
Finding

The argparse description, argument help text, and printed status messages are all hard-coded in Chinese. Because these are direct user-facing strings and no language selection or justification is provided, the skill enforces a specific language/locale by default.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
84% confidence
Finding

The natural-language description and instructions are presented entirely in Chinese and do not indicate that users may choose another language. Under the policy, forcing a specific language without opt-in can be a locale/language policy violation unless the regional scope is explicitly justified.

Content

No source excerpt is available for this finding.

Missing User Warnings

Low
Category
Not specified by scanner
Confidence
80% confidence
Finding

Automatically sourcing and embedding external images without warning introduces provenance, privacy, and copyright risk. Even if not a direct system-compromise vector, users may unknowingly redistribute copyrighted or misleading images in generated presentations.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Low
Category
Not specified by scanner
Confidence
95% confidence
Finding

The module docstring says it '生成旅行攻略 PPT' and '调用 powerpoint-pptx skill 创建精美 PPT', implying actual PPT generation and delegation to another skill. In the code, there is no invocation of any external PPT-generating skill or library; it only assembles slide data and prints or writes JSON content.

Content

No source excerpt is available for this finding.

Intent-Code Divergence

Low
Category
Not specified by scanner
Confidence
97% confidence
Finding

This is an intent-code divergence in the file's own documentation: the docstring explicitly describes calling another skill to create the PPT. However, the executable logic only constructs slide dictionaries, serializes them to JSON, and optionally writes that JSON to disk.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
95% confidence
Finding

This code primarily generates Chinese-language PPT content, but it inserts the English phrase "discover the best of {city}" directly into the slide content. The file also includes another fixed English phrase later, which indicates a language choice is being imposed rather than offered to the user.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
93% confidence
Finding

The closing slide content is otherwise in Chinese but appends the English phrase "Safe Travels ✈️" unconditionally. This can violate language/locale policy because the script does not provide a documented language selection or explain why mixed-language output is required.

Content

No source excerpt is available for this finding.

Intent-Code Divergence

Low
Category
Not specified by scanner
Confidence
78% confidence
Finding

Within this file, 'generate_ppt_data' only assembles a dictionary describing slide content and does not create a presentation file or fetch images. Because the surrounding skill promise is to automatically generate a polished PPT, the in-code documentation can mislead reviewers into thinking PPT generation is implemented here when it is only a metadata scaffold.

Content

No source excerpt is available for this finding.