Audio Cog
Security checks across malware telemetry and agentic risk
Overview
Audio Cog is a coherent CellCog audio-generation integration, but it openly requires a CellCog API key and sends work to external/async voice services, including voice cloning.
Install this skill only if you trust CellCog and are comfortable sending prompts and audio-generation requests to its service. Use a controlled API key, monitor usage, avoid submitting secrets, and use cloned voices only with consent and clear disclosure.
VirusTotal
66/66 vendors flagged this skill as clean.
Risk analysis
Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.
Using the skill may consume CellCog account resources or access account-associated audio-generation features.
The skill requires a CellCog API key, which is expected for the stated service integration but grants access to the user's CellCog account or billing context.
env: [CELLCOG_API_KEY]
Use a dedicated or least-privileged CellCog API key if available, monitor usage, and avoid sharing the key in prompts or generated content.
Prompt text, task labels, and audio-generation requests may leave the local agent and be processed by CellCog or its voice providers.
The documented workflow sends user prompts into a CellCog chat/agent task and reports back through a session key, indicating an external provider/agent communication flow.
result = client.create_chat(
prompt="[your task prompt]",
notify_session_key="agent:main:main",
task_label="my-task",
chat_mode="agent",
)Do not include secrets or private text unless you intend to send it to CellCog, and review CellCog's privacy and retention policies before use.
If the environment installs or resolves this dependency, trust shifts to the external CellCog package/source.
The artifact relies on an external CellCog dependency, but no install spec or package pin is provided in the supplied artifacts.
dependencies: [cellcog]
Install CellCog only from an official source, prefer pinned versions where possible, and review the companion CellCog skill/package if available.
An audio-generation job may continue running after the initial call and could consume service credits or produce output later.
The documented OpenClaw workflow intentionally starts an asynchronous task rather than blocking until completion.
**OpenClaw (fire-and-forget):**
Use clear task labels, avoid launching duplicate long-running jobs, and check CellCog's status or cancellation controls for larger tasks.
Generated audio could be mistaken for a real person's speech if shared without disclosure.
The skill explicitly supports cloned/avatar voices, a disclosed and purpose-aligned capability that can affect human trust if used without consent or labeling.
When an avatar has a cloned voice, CellCog uses the MiniMax provider to generate speech that sounds like that person.
Use voice cloning only with appropriate consent and label synthetic or cloned-voice audio when sharing it.