Skillv1.0.2

ClawScan security

Elevenlabs Toolkit · ClawHub's context-aware review of the artifact, metadata, and declared behavior.

Scanner verdict

ReviewMar 28, 2026, 4:40 AM

Verdict: Review
Confidence: medium
Model: gpt-5-mini
Summary: The skill largely matches its ElevenLabs TTS/STT purpose, but it includes undeclared external dependencies and accesses an additional API key (MISTRAL_API_KEY) not listed in requirements — these inconsistencies merit caution.
Guidance: This skill appears to implement the ElevenLabs features it advertises, but you should be cautious before installing or running it: 1) The package includes Python code that requires additional libraries (fastapi, httpx, websockets, mistralai, etc.) but provides no install instructions — ask the author for a requirements file or installation spec or prepare to install dependencies yourself. 2) The code can optionally call Mistral if MISTRAL_API_KEY is present, but that env var is not declared; if you do not want it to call Mistral, ensure MISTRAL_API_KEY is not set in your environment. 3) The skill needs outbound network access and your ELEVENLABS_API_KEY; never share that key with untrusted code. 4) Confirm expected behavior (for streaming, STT uploads, and conversational features) in a safe environment before using in production. If you need higher assurance, request the author to: (a) declare all required env vars (including optional ones), (b) provide a requirements.txt or install spec that uses trusted package sources, and (c) document exactly when additional services (like Mistral) will be invoked.

Review Dimensions

Purpose & Capability: okThe name/description (ElevenLabs TTS, STT, SFX, music, streaming, voice isolation) align with the code and SKILL.md: the code proxies to api.elevenlabs.io endpoints for voices, text-to-speech, sound generation, speech-to-text, isolation, and streaming. ELEVENLABS_API_KEY is declared and used as the primary credential.
Instruction Scope: noteSKILL.md and the included Python implement only the declared ElevenLabs features and expose FastAPI endpoints for them. However, the code also implements a conversational 'story concierge' that calls a third-party Mistral client if MISTRAL_API_KEY is present — this behavior is not declared in requires.env and broadens the runtime scope. SKILL.md's metadata mentions base64 usage but the implementation returns raw bytes (minor inconsistency).
Install Mechanism: concernThere is no install spec, yet the included code depends on multiple Python packages (fastapi, starlette, httpx, websockets, mistralai, etc.). Without a declared install step, an environment running this skill may lack required dependencies or the operator may need to install them manually; that absence is an operational and supply-chain mismatch (not necessarily malicious but worth noting).
Credentials: concernELEVENLABS_API_KEY is appropriate and declared as primary. The code optionally reads MISTRAL_API_KEY and imports a 'mistralai' client to call another service, but MISTRAL_API_KEY is not listed in requires.env. Requesting or using additional service credentials without declaration is a proportionality/information-gap concern.
Persistence & Privilege: okThe skill does not request always:true, does not modify other skills, and has no declared persistent/system-level privileges. It performs outbound network calls to ElevenLabs (expected for the stated purpose).