Waxell

skill

asr-transcribe-to-text High risk

daymadedaymade/claude-code-skillsupdated

Transcribes audio or video to speaker-labeled, timestamped text, locally with MLX on Apple Silicon or remotely. Use for 转录 / 录音转文字 / 说话人分离 / 字幕, and also for preparing audio for ASR without transcribing: 转格式, 降采样到 16kHz, merging recorder segments, or compressing and speeding up audio before 飞书妙记, even when it looks like a one-line ffmpeg job.
daymade's own description, from the skill's files.
Stars
1,534
Version
v2
Copies
0

Overview

blockHigh risk

Waxell rates it high risk: 50 findings, highest severity critical.

Highest severity
critical
Findings
50
What ran
Static analysis + deep scan

A clean result means nothing alarming in what is published, not a guarantee. How the checks work

We don't republish the skill's instructions. Read them on GitHub at the reviewed commit.

Security

Security scan

Checked yesterday

Result

High risk Waxell rates it high risk: 50 findings, highest severity critical.

Checked with static analysis + deep scan: servers it launches, packages it installs at runtime, scripts piped to a shell and bundled binaries. The deep scan also read the instructions for prompt injection and data exfiltration. Nothing was run. How the checks work

Findings (50)

critical · 2

  • critical

    Data Flow: Tainted flow: 'request' from os.environ.get (line 205, credential/environment) → urllib.request.urlopen (netw

    scripts/voiceprint_id.py:206 · skillspector:TT3

  • critical

    Dangerous Code Execution: Dangerous chain: exec() wrapping compile

    tests/test_execution_evidence.py:112 · skillspector:AST8

high · 15

  • high

    analysis-evasion: Incomplete referenced artifact analysis

    SKILL.md:512 · skillspector:AE1

  • high

    analysis-evasion: Incomplete referenced artifact analysis

    SKILL.md:1052 · skillspector:AE1

  • high

    Tool Misuse: Tool Parameter Abuse

    SKILL.md:838 · skillspector:TM1

  • high

    analysis-evasion: Incomplete referenced artifact analysis

    SKILL.md:1044 · skillspector:AE1

  • high

    Memory Poisoning: Memory Manipulation

    references/review_workbench.md:134 · skillspector:MP3

  • high

    Dangerous Code Execution: os.system() or os exec-family call

    scripts/external_dispatch.py:203 · skillspector:AST5

  • high

    Data Exfiltration: Env Variable Harvesting

    scripts/external_dispatch.py:203 · skillspector:E2

  • high

    Output Handling: Unvalidated Output Injection

    scripts/review-workbench/tests/test_prepare_project.py:302 · skillspector:OH1

  • high

    Data Exfiltration: Env Variable Harvesting

    tests/test_dispatch_resume.py:156 · skillspector:E2

  • high

    Data Exfiltration: Env Variable Harvesting

    tests/test_dispatch_resume.py:194 · skillspector:E2

  • high

    Dangerous Code Execution: exec() call detected

    tests/test_execution_evidence.py:112 · skillspector:AST1

  • high

    Data Exfiltration: Env Variable Harvesting

    tests/test_external_dispatch.py:54 · skillspector:E2

  • high

    Output Handling: Unvalidated Output Injection

    tests/test_external_dispatch.py:174 · skillspector:OH1

  • high

    Data Exfiltration: Env Variable Harvesting

    tests/test_voiceprint_bank.py:151 · skillspector:E2

  • high

    Data Exfiltration: Env Variable Harvesting

    tests/test_voiceprint_bank.py:176 · skillspector:E2

medium · 33

  • medium

    Suspicious Unicode normalization or mixed-script content

    SKILL.md:1 · skillspector:AE4

  • medium

    Without declared permissions the skill's intent is opaque and cannot be validated.

    SKILL.md:1 · skillspector:LP3

  • medium

    Excessive Agency: Autonomous Decision Making

    SKILL.md:832 · skillspector:EA2

  • medium

    Dangerous Code Execution: subprocess module call

    scripts/build_whispercpp_gpu.py:33 · skillspector:AST4

  • medium

    Dangerous Code Execution: subprocess module call

    scripts/build_whispercpp_gpu.py:38 · skillspector:AST4

  • medium

    Dangerous Code Execution: subprocess module call

    scripts/measure_asr.py:84 · skillspector:AST4

  • medium

    Dangerous Code Execution: subprocess module call

    scripts/overlap_merge_transcribe.py:26 · skillspector:AST4

  • medium

    Dangerous Code Execution: subprocess module call

    scripts/overlap_merge_transcribe.py:57 · skillspector:AST4

  • medium

    Dangerous Code Execution: subprocess module call

    scripts/overlap_merge_transcribe.py:82 · skillspector:AST4

  • medium

    Dangerous Code Execution: subprocess module call

    scripts/prepare_asr_input.py:53 · skillspector:AST4

  • medium

    Dangerous Code Execution: subprocess module call

    scripts/resolve_media_input.py:277 · skillspector:AST4

  • medium

    Dangerous Code Execution: subprocess module call

    scripts/resolve_media_input.py:303 · skillspector:AST4

  • medium

    Dangerous Code Execution: subprocess module call

    scripts/review-workbench/prepare_project.py:108 · skillspector:AST4

  • medium

    Dangerous Code Execution: subprocess module call

    scripts/review-workbench/prepare_project.py:183 · skillspector:AST4

  • medium

    Rogue Agent: Session Persistence

    scripts/review-workbench/server.py:622 · skillspector:RA2

  • medium

    Dangerous Code Execution: subprocess module call

    scripts/review-workbench/tests/test_prepare_project.py:254 · skillspector:AST4

  • medium

    Dangerous Code Execution: subprocess module call

    scripts/review-workbench/tests/test_prepare_project.py:302 · skillspector:AST4

  • medium

    Excessive Agency: Unbounded Resource Access

    scripts/speaker_transcribe.py:595 · skillspector:EA4

  • medium

    Excessive Agency: Unbounded Resource Access

    tests/test_long_audio_safety.py:783 · skillspector:EA4

  • medium

    Excessive Agency: Unbounded Resource Access

    tests/test_long_audio_safety.py:1064 · skillspector:EA4

  • medium

    Excessive Agency: Unbounded Resource Access

    tests/test_long_audio_safety.py:1110 · skillspector:EA4

  • medium

    Dangerous Code Execution: subprocess module call

    scripts/speaker_transcribe.py:597 · skillspector:AST4

  • medium

    Dangerous Code Execution: subprocess module call

    scripts/speaker_transcribe_cascade.py:116 · skillspector:AST4

  • medium

    Dangerous Code Execution: subprocess module call

    scripts/speaker_transcribe_cascade.py:125 · skillspector:AST4

  • medium

    Dangerous Code Execution: subprocess module call

    scripts/transcribe_long_whispercpp.py:513 · skillspector:AST4

  • medium

    Excessive Agency: Unbounded Resource Access

    scripts/transcribe_long_whispercpp.py:532 · skillspector:EA4

  • medium

    Excessive Agency: Unbounded Resource Access

    tests/test_long_audio_safety.py:698 · skillspector:EA4

  • medium

    Dangerous Code Execution: subprocess module call

    scripts/transcribe_remote.py:52 · skillspector:AST4

  • medium

    Rogue Agent: Session Persistence

    scripts/voiceprint_id.py:15 · skillspector:RA2

  • medium

    Dangerous Code Execution: subprocess module call

    scripts/voiceprint_id.py:83 · skillspector:AST4

  • medium

    Dangerous Code Execution: subprocess module call

    scripts/voiceprint_id.py:232 · skillspector:AST4

  • medium

    Data Flow: Tainted flow: 'command' from os.environ.get (line 229, credential/environment) → subprocess.run (code executi

    scripts/voiceprint_id.py:232 · skillspector:TT2

  • medium

    Dangerous Code Execution: subprocess module call

    tests/test_dispatch_resume.py:159 · skillspector:AST4

Versions

Version history

Each distinct content we have seen, newest first, with the verdict it got at the time.

  1. v22026-11-166 added, 9 changed High riskgot riskier
  2. v12026-11-03 No issues

Files

Files

57 files, 1.3 MB in total. Names only; open them on GitHub.

  • .gitignore
  • SKILL.md
  • references/batch_vad_coverage.md
  • references/decoupled_speaker_alignment.md
  • references/execution_location_and_comparison.md
  • references/local_mlx_guide.md
  • references/overlap_merge_strategy.md
  • references/review_workbench.md
  • references/speaker_diarization.md
  • references/voiceprint_speaker_id.md
  • references/whisper_word_timestamps.md
  • references/whispercpp_gpu_runtime.md
  • scripts/align_speakers.py
  • scripts/build_whispercpp_gpu.py
  • scripts/diarize_speakers.py
  • scripts/diarize_speakers.py.lock
  • scripts/dispatch_compat.py
  • scripts/dispatch_compatibility.json
  • scripts/external_dispatch.py
  • scripts/fuse_whispercpp_diarization.py
  • scripts/generate_audit_html.py
  • scripts/measure_asr.py
  • scripts/overlap_merge_transcribe.py
  • scripts/prepare_asr_input.py
  • scripts/resolve_media_input.py
  • scripts/review-workbench/prepare_project.py
  • scripts/review-workbench/server.py
  • scripts/review-workbench/static/app.js
  • scripts/review-workbench/static/index.html
  • scripts/review-workbench/static/styles.css
  • scripts/review-workbench/tests/test_client_integrity.cjs
  • scripts/review-workbench/tests/test_frontend.cjs
  • scripts/review-workbench/tests/test_playback.cjs
  • scripts/review-workbench/tests/test_prepare_project.py
  • scripts/review-workbench/tests/test_server.py
  • scripts/review-workbench/tests/test_server_integrity.py
  • scripts/speaker_time_contract.py
  • scripts/speaker_transcribe.py
  • scripts/speaker_transcribe.py.lock
  • scripts/speaker_transcribe_cascade.py
  • scripts/transcribe_local_mlx.py
  • scripts/transcribe_long_whispercpp.py
  • scripts/transcribe_remote.py
  • scripts/voiceprint_id.py
  • scripts/whispercpp-gpu-only-v1.9.1.patch
  • scripts/word_timestamps_whisper.py
  • tests/test_alignment_turns.py
  • tests/test_dispatch_resume.py
  • tests/test_execution_evidence.py
  • tests/test_external_dispatch.py
  • tests/test_long_audio_safety.py
  • tests/test_skill_delivery_contract.py
  • tests/test_speaker_bundle_time_contract.py
  • tests/test_voiceprint_bank.py
  • tests/test_whispercpp_fusion.py
  • tests/test_whispercpp_gpu.py
  • tests/test_whispercpp_long_audio.py

See and control which skills your people use

Inside your organization, Waxell scans each skill and plugin before anyone installs it, approves what passes and publishes it to a private marketplace, and puts every MCP tool call behind one governed URL.

30 minutes. No commitment.

SOC 2 Type II attested · ISO/IEC 27001:2022 certified · HIPAA with a BAA · aligned to the EU AI Act and NIST AI RMF

More images, video & audio skills and plugins

See all