"dataify-tiktok-comment-by-url"

• Outcome: Review
• Latest audit: Sep 1, 2026 · 9:32 AM UTC
• Version: 1.3.0

Overview

This skill is labeled as TikTok comment collection, but its instructions and catalog also enable broader TikTok shop, post-list, and profile collection through Dataify.

Review before installing. This does not show clear malicious behavior or hidden destruction, but it is broader than a TikTok-comments-only skill: it can prepare and submit Dataify scraping tasks for shops, post lists, and profiles, and it uses your Dataify API token for paid external API calls. Install only if you intend to grant that broader Dataify TikTok builder authority, and prefer a narrowed version that allows only tiktok_comment_by-url.

SkillSpector

By NVIDIA

Vulnerability Patterns

• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• Trigger Abuse: Overly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
• MCP Tool Poisoning: Hidden Instructions, Unicode Deception, Parameter Description Injection
• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands
• Privilege Escalation: Excessive Permissions, Sudo/Root Execution, Credential Access

Findings (10)

Tp4

High

• Category: MCP Tool Poisoning
• Confidence: 97% confidence
• Finding: The manifest says the skill is limited to collecting TikTok comments from a known content URL, but the body instructs the agent to select from a broader scraper catalog and submit arbitrary builder requests. That scope expansion weakens user consent and policy boundaries, and could enable collection of unrelated TikTok data through the same skill path.

Description-Behavior Mismatch

High

• Confidence: 99% confidence
• Finding: The workflow explicitly offers shop, posts-by-list URL, and profile collection tools even though the skill description says it must not be used for those purposes. This is a real capability expansion that can drive the agent to collect data outside the declared scope, increasing privacy, compliance, and misuse risk.

Intent-Code Divergence

High

• Confidence: 99% confidence
• Finding: The inline instructions directly contradict the manifest by guiding the agent to use profile and discovery tools that the manifest forbids. Contradictory instructions are dangerous because runtime behavior will likely follow the broader operational steps rather than the narrow declared purpose.

Description-Behavior Mismatch

Medium

• Confidence: 86% confidence
• Finding: The default completion section includes special handling for media downloads, despite the manifest stating the skill must not be used for video download. Even if not directly invoking a downloader here, documenting download-oriented execution paths broadens intended use and can normalize prohibited actions.

Description-Behavior Mismatch

High

• Confidence: 98% confidence
• Finding: The skill metadata says it is limited to collecting TikTok comments by URL, but the workflow expands available tools to shops, posts, and profiles. This scope expansion can cause unauthorized or unintended collection of broader TikTok data than the user or reviewer expects, increasing privacy, compliance, and least-privilege risks.

Intent-Code Divergence

High

• Confidence: 97% confidence
• Finding: The front-matter description claims the skill is rooted in tiktok_comment_by-url, yet it instructs selection among unrelated profile and list-discovery tools. That mismatch undermines transparent consent and can mislead downstream agents into invoking broader scraping actions than the skill name and description authorize.

Description-Behavior Mismatch

Medium

• Confidence: 91% confidence
• Finding: The interface metadata and default prompt broaden the skill’s apparent scope from a narrowly limited 'TikTok comments by URL' action to a generic 'Dataify collection' or 'Tiktok Builder' capability. In an agent setting, this kind of scope mismatch can cause the model or user to invoke the skill for unsupported or overbroad collection tasks, increasing the chance of policy bypass, data overcollection, or misuse beyond the intended constraints.

Description-Behavior Mismatch

Medium

• Confidence: 95% confidence
• Finding: The referenced tool list exposes multiple TikTok collection capabilities beyond the declared skill scope of collecting comments from a known content URL. This creates a scope-expansion risk where the agent could invoke profile, post-list, or shop collection tools contrary to user expectations and policy boundaries, enabling unintended data collection.

Vague Triggers

Medium

• Confidence: 87% confidence
• Finding: The tool descriptors provide only broad URL-oriented labels and empty parameter definitions, with no constraints describing acceptable URL patterns, resource types, or trigger disambiguation. In an agent setting, this ambiguity can cause incorrect tool selection or allow a user-supplied TikTok URL to activate a broader collection capability than intended.

External Transmission

Medium

• Category: Data Exfiltration

Content

13. Set `spider_name` to `tiktok.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

• Confidence: 83% confidence
• Finding: The skill instructs the agent to send collected parameters and an authorization token to an external API endpoint. External transmission is expected for an integration skill, but it still creates real data exposure risk because user-supplied URLs and scraping parameters are sent off-platform and the skill's broadened scope increases the sensitivity of what may be transmitted.

VirusTotal

VirusTotal findings are pending for this skill version.

Static analysis

No suspicious patterns detected.