Convert text into high-quality speech with a macOS say-like workflow.
Overall risk appears low: the material describes a text-to-speech tool for ElevenLabs, and the stated source is an auditable GitHub open-source project with very high community adoption. However, the declaration of “no keys / no remote endpoint” conflicts with the README, which clearly requires an ElevenLabs API key and implies third-party network access, so the real configuration and data flow should be verified before use.
The README explicitly requires `ELEVENLABS_API_KEY` and also supports `SAG_API_KEY`, which conflicts with the earlier claim of “no keys.” This is a sensitive API credential; if exposed via environment variables, logs, or shell history, it could enable abuse of the associated ElevenLabs account.
Although the metadata says “no remote endpoint,” the described functionality is “ElevenLabs text-to-speech,” so it is reasonable to infer that input text is sent to ElevenLabs and audio is returned. The material does not specify exact domains, retention policy, or telemetry behavior, so this should be treated as normal data egress to the declared third-party service.
The system flags this as `prompt-only`, and the material is primarily usage documentation and prompt-like guidance, with no evidence that it independently acquires extra system privileges or executes arbitrary code. The `sag ...` commands in the README are examples of explicit user invocation rather than proof of hidden execution capability.
The material shows local playback and the ability to write an audio file via `-o /tmp/voice-reply.mp3`, indicating at least limited local file output; the text to be synthesized is also sent to a third-party TTS service. There is no stated ability to broadly read local files, access credential stores, or request data beyond the TTS purpose, so this fits a normal caution level rather than over-privileged access.
Positive signals include a GitHub open-source repository and extremely high community adoption (about 377k stars), both of which materially reduce supply-chain risk. Remaining concerns are that the license is undeclared, maintenance status is unknown, and the mapping between the tool name `sag` and the provided repository `openclaw/openclaw` cannot be verified from the material alone, so repository ownership and source-to-release consistency should be confirmed before installation.
Copy the install command and let the AI configure it · recommended for beginners
Please install the "sag" skill from askskill: 1. Download https://raw.githubusercontent.com/openclaw/openclaw/main/skills/sag/SKILL.md 2. Save it as ~/.claude/skills/sag/SKILL.md 3. Reload skills and tell me it's ready
Please read the following article in a natural, clear female voice in Chinese and output an audio file ready for publishing: {article content}A smooth narrated audio file suitable for publishing or previewing content.
Turn this product introduction copy into an English male voiceover with a professional, concise tone for a demo video: {product copy}A professional English product voiceover ready for demo videos or promotional materials.
Convert the following reminder text into a short, clear spoken notification with a slightly faster pace: {reminder text}A concise spoken reminder suitable for notifications, announcements, or accessibility use.
Use sag for ElevenLabs TTS with local playback.
API key (required)
ELEVENLABS_API_KEY (preferred)SAG_API_KEY also supported by the CLIQuick start
sag "Hello there"sag speak -v "Roger" "Hello"sag voicessag prompting (model-specific tips)Model notes
eleven_v3 (expressive)eleven_multilingual_v2eleven_flash_v2_5Pronunciation + delivery rules
--normalize auto (or off if it harms names).--lang en|de|fr|... to guide normalization.<break> not supported; use [pause], [short pause], [long pause].<break time="1.5s" /> supported; <phoneme> not exposed in sag.v3 audio tags (put at the entrance of a line)
[whispers], [shouts], [sings][laughs], [starts laughing], [sighs], [exhales][sarcastic], [curious], [excited], [crying], [mischievously]sag "[whispers] keep this quiet. [short pause] ok?"Voice defaults
ELEVENLABS_VOICE_ID or SAG_VOICE_IDConfirm voice + speaker before long output.
When the user asks for a "voice" reply (e.g., "crazy scientist voice", "explain in voice"), generate audio and send it:
# Generate audio file
sag -v Clawd -o /tmp/voice-reply.mp3 "Your message here"
# Then include in reply:
# MEDIA:/tmp/voice-reply.mp3
Voice character tips:
[excited] tags, dramatic pauses [short pause], vary intensity[whispers] or slower pacing[sings] or [shouts] sparinglyDefault voice for Clawd: lj2rcrvANS3gaWWnczSX (or just -v Clawd)
Verify an OpenClaw release is fully published and working across all channels.
Fetch GitHub issues, create fixes, open PRs, and handle reviews.
Convert text to speech locally and offline with sherpa-onnx, no cloud needed.
Regenerate OpenClaw release changelog sections from Git history before releases.
Prepare and verify OpenClaw stable or beta releases and release notes.
Create and review technical docs and agent instruction files in repositories.
Convert text to speech quickly and browse available voices and models.
Access text-to-speech, speech-to-text, and voice management via ElevenLabs API.
Generate speech, clone voices, transcribe audio, and create sound effects.
Automate ElevenLabs text-to-speech, voice management, and audio history workflows.
Use ElevenLabs TTS and audio processing on Windows with Cursor integration.
Convert text into multilingual speech and save it as MP3 audio.