ClawHub
Back to skill

Security audit

Deepgram Api

Security checks across malware telemetry and agentic risk

Overview

This is a coherent Deepgram documentation skill, but its MCP setup can persistently connect coding tools to Deepgram services with broad tool/account access and limited privacy warning.

Install this only if you are comfortable giving your coding tool Deepgram MCP access. Prefer project-scoped configuration for team repositories, review whether the hosted MCP is allowed for sensitive code or prompts, and use the local CLI option only when you understand what Deepgram account actions the MCP tools can perform.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • 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

Medium
Confidence
93% confidence
Finding
The callback parameter allows Deepgram to send data to an arbitrary user-specified URL, which can expose synthesized job metadata or trigger requests to unintended destinations if integrators use untrusted input for the callback value. The documentation omits warnings about validating callback destinations, protecting secrets in callback URLs, and understanding the privacy implications of sending results to third-party endpoints.

Vague Triggers

Medium
Confidence
91% confidence
Finding
The skill description is broadly phrased to trigger on generic documentation, help-location, and question-answering requests about Deepgram across many product areas. Overly broad activation can cause the skill to activate in contexts where it was not specifically intended, increasing the chance of prompt hijacking, irrelevant tool invocation, or untrusted documentation being surfaced in sensitive workflows.

Missing User Warnings

Medium
Confidence
94% confidence
Finding
The hosted MCP instructions direct users to connect their coding tool to a remote HTTP endpoint without clearly warning that prompts, documentation queries, and possibly surrounding context may be transmitted off-host to a third-party service. In an AI coding tool context, this can expose sensitive project data if users assume the integration is local or privacy-equivalent to the CLI MCP option.

External Transmission

Medium
Category
Data Exfiltration
Content
"mcpServers": {
    "deepgram-docs": {
      "type": "http",
      "url": "https://api.dx.deepgram.com/kapa/mcp"
    }
  }
}
Confidence
93% confidence
Finding
https://api.dx.deepgram.com/

External Transmission

Medium
Category
Data Exfiltration
Content
"mcpServers": {
    "deepgram-docs": {
      "type": "http",
      "url": "https://api.dx.deepgram.com/kapa/mcp"
    }
  }
}
Confidence
93% confidence
Finding
https://api.dx.deepgram.com/

External Transmission

Medium
Category
Data Exfiltration
Content
Provide the MCP server details and let them configure manually:

- **Type:** HTTP
- **URL:** `https://api.dx.deepgram.com/kapa/mcp`

---
Confidence
92% confidence
Finding
https://api.dx.deepgram.com/

VirusTotal

VirusTotal findings are pending for this skill version.

View on VirusTotal

Static analysis

No suspicious patterns detected.