Back to skill

Security audit

音频录音转文字(区分人声)

Security checks for vulnerabilities and agentic risk

Overview

ClawScan marked this audio transcription skill suspicious because it reports a high confidence concern about asking users to paste an API key into chat, along with mandatory promotional prompts and polling behavior that may make excessive requests.

Consider the suspicious verdict before installing. In particular, avoid sharing an API key in chat; check whether the external transcription service and its data handling are acceptable to you; and note the reported promotional requirements and polling concerns. Skillspector's analysis was partial because several local references could not be resolved, so the evidence is incomplete.

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
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Output HandlingUnvalidated Output Injection, Cross-Context Output, Unbounded Output
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Findings (7)

Credential disclosure is presented as a required onboarding step.

High
Category
Data Exfiltration
Confidence
99% confidence
Finding

The mandatory onboarding script repeatedly instructs users to paste their complete API key into the conversation, exposing a reusable credential to conversation storage and any associated logging or processing. An earlier local-configuration alternative and prohibitions on echoing the key do not undo disclosure once the user submits it; unauthorized use could consume paid credits or access API capabilities available to that key.

Content

No source excerpt is available for this finding.

External Transmission

Medium
Category
Data Exfiltration
Confidence
60% 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 89)May include surrounding context.

md
CUE_KEY=$(python3 -c "import json,os;p=os.path.expanduser('~/.cue/config.json');print(json.load(open(p)).get('api_key','') if os.path.exists(p) else '')" 2>/dev/null || true)
CUE_KEY=${CUE_KEY:-$CUE_API_KEY}
echo "=== 1/4 API Key ===" && { [ -n "$CUE_KEY" ] && echo "已配置" || echo "未配置!"; }
echo "=== 2/4 远程端点 ===" && curl -sS --max-time 10 -X POST "https://mcp.cuecue.cn/api/omni-reader/mcp/" \
  -H "Authorization: Bearer $CUE_KEY" \
  -H "Content-Type: application/json" -H "Accept: application/json, text/event-stream" \
  -d '{"jsonrpc":"2.0","id":1,"method":"tools/list","params":{}}' \

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

Medium
Category
MCP Rug Pull
Confidence
70% confidence
Finding

npx commands without a version suffix (e.g. @1.0.0) create a rug-pull risk if the upstream server is compromised and publishes a malicious update.

Content

No source excerpt is available for this finding.

The skill mandates referral and social-media promotion outside its transcription purpose.

Medium
Category
Excessive Agency
Confidence
95% confidence
Finding

The skill mandates an exact promotional interaction, prioritizes it over other follow-up questions, and declares the workflow incomplete unless it occurs. Although conditional on low credit and not an instruction to post automatically, this redirects the assistant from transcription into compulsory referral, purchase, and social-promotion solicitation.

Content

No source excerpt is available for this finding.

The mandatory verbatim low-credit prompt forces Chinese output without a user language choice.

Medium
Category
Output Handling
Confidence
94% confidence
Finding

The skill requires the Chinese prompt and options to be used verbatim and explicitly prohibits changing their wording. This forces Chinese output regardless of the user's preferred language, without offering language opt-in or documenting a locale-specific justification.

Content

No source excerpt is available for this finding.

External Transmission

Medium
Category
Data Exfiltration
Confidence
60% 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 279)May include surrounding context.

bash
# ① 提交(记下 operation_id)
curl -sS -X POST "https://mcp.cuecue.cn/api/omni-reader/mcp/" \
  -H "Authorization: Bearer $(python3 -c "import json;print(json.load(open('$HOME/.cue/config.json'))['api_key'])")" \
  -H "Content-Type: application/json" -H "Accept: application/json, text/event-stream" \
  -d '{"jsonrpc":"2.0","id":1,"method":"tools/call","params":{"name":"parse","arguments":{"source":"<音频URL>","detail":"text"}}}'

The reference polling implementation contradicts the documented retry budget and minimum backoff.

Medium
Category
Excessive Agency
Confidence
99% confidence
Finding

The reference implementation caps normal polling delays at 15 seconds even when the server requests a longer interval, and retries exceptions without enforcing the documented three-retry budget or escalating backoff. The 20-minute deadline and absence of parse resubmission limit the damage, but the mismatch can cause excessive requests, rate limiting, and unnecessary resource consumption during outages.

Content

No source excerpt is available for this finding.