ClawHub
Skill flagged — suspicious patterns detected

ClawHub Security flagged this skill as suspicious. Review the scan results before using.

Text to speech using the default macOS "say" command. No need for 3rd party APIs or models. Supports many languages. Also, Trinoids!

v0.0.2

Local text-to-speech using macOS `say` + ffmpeg for Telegram/Matrix voice messages

⭐ 0· 16·0 current·0 all-time
by@zviratko
MIT-0
Download zip
LicenseMIT-0 · Free to use, modify, and redistribute. No attribution required.
Security Scan
VirusTotalVirusTotal
Suspicious
View report →
OpenClawOpenClaw
Benign
high confidence
ℹ
Purpose & Capability
The name/description match the actual instructions: generating AIFF with macOS `say` then converting to Opus with `ffmpeg` is exactly what's needed for Telegram/Matrix voice messages. One minor inconsistency: skill metadata lists no OS restriction even though `say` is macOS-specific; the declared required binaries (`say`, `ffmpeg`) properly reflect the true platform dependency.
ℹ
Instruction Scope
SKILL.md stays on‑topic (create AIFF with `say`, transcode with `ffmpeg`, then send with the message tool). No unrelated files, credentials, or external endpoints are referenced. Caution: the examples use shell interpolation (VOICE/TEXT variables). If untrusted input is passed into shell commands without proper escaping, there is a risk of shell/command injection — ensure the agent or runtime invokes `say`/`ffmpeg` with safely escaped arguments or argument lists rather than raw shell interpolation.
✓
Install Mechanism
No install spec (instruction-only) — minimal risk because nothing is downloaded or written by the skill itself. The runtime relies on system-installed `say` and `ffmpeg`.
✓
Credentials
No environment variables, credentials, or config paths are requested. The skill does suggest a workspace path (~/.openclaw/workspace/tmp/audio) for temporary files; this is reasonable but should be whitelisted only if you accept exec access for that limited path.
✓
Persistence & Privilege
always is false and the skill is user-invocable; it does not request permanent or elevated presence. The only privilege-related suggestion is to whitelist a dedicated workspace path for exec permissions to speed approvals — keep permissions scoped to that path.
Assessment
What to consider before installing: 1) This only works when `say` exists (macOS); install `ffmpeg` and test the `say`→`ffmpeg` pipeline locally first. 2) Approve exec permissions only for a narrow workspace path (e.g., ~/.openclaw/workspace/tmp/audio). 3) Ensure the runtime escapes user-provided TEXT/VOICE variables (avoid passing raw, unsanitized strings to a shell) to prevent command injection. 4) Confirm how the 'message' tool is authorized to send media and avoid sending sensitive audio unintentionally. 5) If you run on non-macOS hosts, the skill will fail (metadata omits an OS restriction but the required binary `say` is macOS-only).

Like a lobster shell, security has layers — review code before you run it.

latestvk978hzf9hcba7eqfh71q4r2hs984p5wc

License

MIT-0
Free to use, modify, and redistribute. No attribution required.
Termshttps://spdx.org/licenses/MIT-0.html

Runtime requirements

🎤 Clawdis
Binssay, ffmpeg

Comments