Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add CocoRoF/Geny --skill whiteboard_voice_notesgit clone --depth 1 https://github.com/CocoRoF/GenyWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/cocorof/geny/whiteboard_voice_notes)<a href="https://agentmods.dev/skills/cocorof/geny/whiteboard_voice_notes"><img src="https://agentmods.dev/badge/skills/cocorof/geny/whiteboard_voice_notes.svg" alt="Measured on agentmods" height="20"></a>- NVIDIA SkillSpector pass
What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00059 | $0.01800 |
| Opus 5 | $0.00030 | $0.00900 |
| Sonnet 5 | $0.00012 | $0.00360 |
| Haiku 4.5 | $0.00006 | $0.00180 |
Grade A, and why
whiteboard-voice-notes scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 7d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 128 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Whiteboard: Voice Notes — React to voice notes and re-transcribe when needed
Use this skill when the user shares a microphone recording (microphone_record capture) or an audio file (.webm/.mp3/.m4a/...).
⚠️ Most important: distinguish ambient vs deliberate
You must check the ambient field of the [USER_SHARED] payload (or share_source == "vtuber_stt_stream" / metadata.source == "vtuber_stt_stream"). The two cases mean completely different things:
A. ambient: true (= speech the STT mode happened to pick up)
- The user did not speak directly to you. The mic just overheard a remark nearby.
- It could be the user talking to themselves / a conversation with someone else / singing / TV sound / cursing.
- Default behavior: stay silent. Don't respond; just accumulate the spotlight context.
- Respond only when one of these three holds:
- Your own name was called (the persona name / your nickname).
- A clear direct question to you (ends in
?and points at you). - A clear reaction to your previous statement ("yeah, right", "no, that's not it", etc.).
- If you do need to respond, keep it to a short 1–2 sentence acknowledgment. Use an overheard tone like "I just heard you say [content]..." / "I heard [X] over there". Never use phrasings like "thanks for sharing" / "the memo you sent me".
- When several ambient items are sitting in the spotlight at the same time within one burst, look at the whole thing and react just once. Don't reply to each utterance.
B. ambient: false (= the user shared deliberately)
- The user recorded a memo and sent it (
microphone_record+ Share-with-VTuber), pressed the Share button, or explicitly shared a note. - React as usual. Tone: "I listened to the memo you just recorded" / "Looking at the [X] you shared…".
Background knowledge: auto-transcription has already run
The W2 PostCaptureHook automatically invokes Whisper-large-v3 on every type=audio capture. The result is placed as a quote block at the very top of the note body:
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 7d ago First seen · 128 lines · 59 tokens per session scan A 2edd3b745223
whiteboard-voice-notes is a skill published in the GitHub repository CocoRoF/Geny (10 stars, last pushed 9d ago), licensed Apache-2.0. It adds 59 tokens to every session and 1,800 once invoked, about $0.0003 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other skills, from other repositories
template-engine
Load, validate, and populate templates consistently across frameworks.
checkpoint
Create, list, or recover mid-workflow checkpoints so interrupted work resumes from a known-good position.
update
Update AIWG through its install-aware updater and re-deploy all installed frameworks from the registry.
run
Execute a named script defined in .aiwg/aiwg.config or list all available scripts.
runtime-info
Display a comprehensive runtime environment report covering detected provider, available tools, AIWG version, and installed frameworks.
uninstall-plugin
Remove a previously installed Claude Code marketplace plugin and unregister it from the local plugin registry.