Back to skill

Security audit

视频字幕转文字 okdata

Security checks for vulnerabilities and agentic risk

Overview

The skill is rated suspicious because its transcript workflow includes a fallback example that instead requests unrelated Weibo trends; the evidence is incomplete because reference checks were partial.

Review the fallback example before installing or using the skill. The scan reports that it could make an unrelated authenticated request and possibly incur a charge. Treat the assessment as incomplete because the scanner could not resolve several local references.

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 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 (3)

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
100% confidence
Finding

The HTTP fallback example targets Weibo trends rather than the requested YouTube transcript. An agent copying it literally could make an unintended authenticated, potentially billable API request; however, the surrounding instructions identify the correct endpoint and explain URL construction, making this a limited workflow-integrity issue rather than evidence of malicious behavior.

Content

No source excerpt is available for this finding.

External Transmission

Medium
Category
Data Exfiltration
Confidence
50% confidence
Finding

Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

Content

Scanner excerpt · SKILL.md (reported line 65)May include surrounding context.

go in the query string, and the key comes from the environment, never from the chat:

bash
curl -sG "https://api.okdata.online/v1/weibo/trends/list" \
  -H "Authorization: Bearer $OKDATA_API_KEY" \
  -H "Idempotency-Key: $(uuidgen)" \
  --data-urlencode "view=compact"

The HTTP fallback command fetches Weibo trends instead of a YouTube transcript.

Medium
Category
Excessive Agency
Confidence
98% confidence
Finding

This is the same incorrect fallback example identified by TP4, not an independent vulnerability. Literal execution would retrieve unrelated data using the user's API account and might incur a charge, but it does not demonstrate credential theft, arbitrary execution, or transmission to an unrelated recipient.

Content

No source excerpt is available for this finding.