Back to skill

Security audit

Glyphcall Meeting Transcription

Security checks for vulnerabilities and agentic risk

Overview

ClawScan marked this skill suspicious because it sends audio URLs and text to an external service without a privacy warning, and its examples run an unpinned CLI package.

Before installing, consider whether you are comfortable sending recordings and spoken text to Glyphcall for processing. Avoid sensitive recordings or text unless you have authorization and understand the provider’s handling, and review the use of the changing `agentcash@latest` package.

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
  • Output HandlingUnvalidated Output Injection, Cross-Context Output, Unbounded Output
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (8)

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 25)May include surrounding context.

md
| Task | Endpoint | Price |
|------|----------|-------|
| Speech to text | `https://api.glyphcall.com/transcribe?url=<public-audio-url>` | $0.01 |
| Speaker labels | `https://api.glyphcall.com/transcribe/speakers?...%60= | $0.01 |
| Text to speech | `https://api.glyphcall.com/text-to-speech?text=<text>` | $0.08 |

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 26)May include surrounding context.

TaskEndpointPrice
Speech to texthttps://api.glyphcall.com/transcribe?url=<public-audio-url>$0.01
Speaker labels`https://api.glyphcall.com/transcribe/speakers?...%60=$0.01
Text to speechhttps://api.glyphcall.com/text-to-speech?text=<text>$0.08
sh

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 31)May include surrounding context.

sh
npx agentcash@latest check "https://api.glyphcall.com/transcribe?url=https%3A%2F%2Fexample.com%2Fmeeting.mp3"
npx agentcash@latest fetch "https://api.glyphcall.com/transcribe?url=https%3A%2F%2Fexample.com%2Fmeeting.mp3"

Whisper large-v3-turbo is used; names and numbers in automatic transcripts should be checked.

External Transmission

Medium
Category
Data Exfiltration
Confidence
88% confidence
Finding

External text-to-speech processing is expected, but the documented interface places the full input text in a URL query parameter. Sensitive spoken responses could consequently be retained in URL logs, monitoring systems, or tool traces; HTTPS does not prevent logging at clients and service infrastructure.

Content

Scanner excerpt · SKILL.md (reported line 27)May include surrounding context.

|------|----------|-------| | Speech to text | https://api.glyphcall.com/transcribe?url=<public-audio-url> | $0.01 | | Speaker labels | https://api.glyphcall.com/transcribe/speakers?...%60= | $0.01 | | Text to speech | https://api.glyphcall.com/text-to-speech?text=` | $0.08 |

sh
npx agentcash@latest check "https://api.glyphcall.com/transcribe?url=https%3A%2F%2Fexample.com%2Fmeeting.mp3"

The transcription workflow sends an audio URL to a third-party service without a privacy warning.

Medium
Category
Output Handling
Confidence
96% confidence
Finding

The workflow directs the agent to submit an audio URL to an external processor without explaining that the provider can retrieve and process the recording. Public reachability does not establish consent for third-party processing, particularly for meeting recordings or unlisted capability URLs.

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 30)May include surrounding context.

| Text to speech | https://api.glyphcall.com/text-to-speech?text=<text> | $0.08 |

sh
npx agentcash@latest check "https://api.glyphcall.com/transcribe?url=https%3A%2F%2Fexample.com%2Fmeeting.mp3"
npx agentcash@latest fetch "https://api.glyphcall.com/transcribe?url=https%3A%2F%2Fexample.com%2Fmeeting.mp3"

MCP server referenced without pinned version: 'npx agentcash'.

Medium
Category
MCP Rug Pull
Confidence
98% confidence
Finding

The command uses agentcash@latest, a mutable package reference rather than an exact version. Although this is a CLI invocation rather than evidence of an MCP server, npx can download and execute a newly compromised release with the agent's local permissions.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx agentcash'.

Medium
Category
MCP Rug Pull
Confidence
98% confidence
Finding

The fetch example also executes agentcash@latest without fixing the package version. A malicious or compromised future release could execute code with the agent's permissions; this repeats the same supply-chain weakness as the preceding command.

Content

No source excerpt is available for this finding.