fishaudio-tts

Text-to-Speech using FishAudio (fish.audio), generates natural human-like voice with great emotional expression.

Security Scan

The skill mostly does what it claims (calls FishAudio to produce MP3s) but its metadata and instructions fail to declare that it reads a workspace TOOLS.md and requires an API key, which is an incoherence worth caution.

Purpose & Capability

The description is a FishAudio TTS client, which is coherent with the included Python script. However the registry metadata declares no required environment variables or config paths while the code clearly needs a FishAudio API key (FISH_AUDIO_API_KEY) and also attempts to read ~/.openclaw/workspace/TOOLS.md for an API key. That mismatch (undisclosed credential requirement and file access) is disproportionate to the claimed metadata.

Instruction Scope

SKILL.md tells users to put their API key in an env var or TOOLS.md, but the runtime instructions in the included script explicitly read ~/.openclaw/workspace/TOOLS.md (searching for 'fish'/'api'/'key') and also looks for a specific local proxy string (127.0.0.1:7890). Reading a user workspace file was not declared in metadata and expands the scope of data the skill touches; while the code only parses for a key/proxy, it still reads an undeclared local file.

Install Mechanism

No install spec; this is an instruction-only skill with a single Python script that uses the widely used requests library. There are no remote downloads or archives, and the SKILL.md suggests 'pip install requests' only. Install risk is low.

Credentials

The skill actually requires an API key (FISH_AUDIO_API_KEY) even though the registry lists none. It also reads a local TOOLS.md for the key and proxy configuration. Requiring access to an undeclared user file and a secret is disproportionate to what the metadata claims and should be explicitly declared. The script does not request unrelated cloud creds, but the missing declaration and file read are notable.

Persistence & Privilege

The skill does not request 'always' presence and does not modify other skills or system-wide settings. It writes generated audio files to user-specified paths (expected behavior) and has no special persistence or elevated privileges.

What to consider before installing

This skill appears to be a straightforward FishAudio TTS client, but be aware of two issues before using it: (1) the registry metadata does NOT declare that the script needs an API key, yet the code requires FISH_AUDIO_API_KEY (or will try to extract one from ~/.openclaw/workspace/TOOLS.md). (2) the script reads that specific TOOLS.md file (and checks it for a proxy string), which could contain other secrets you didn't intend to expose. Recommendations: inspect the fish_tts.py source yourself (it is included) and only run it if you trust it; prefer setting the API key via an environment variable rather than storing it in TOOLS.md; verify that ~/.openclaw/workspace/TOOLS.md doesn't contain any other sensitive tokens before allowing the skill to read it; run the script in an isolated environment or sandbox and monitor network requests to ensure it calls the expected https://api.fish.audio endpoint. If you are uncomfortable with undeclared file access, ask the publisher to update the metadata to declare the required env var and config path (or remove implicit TOOLS.md parsing).

High-quality text-to-speech using FishAudio (fish.audio), generates natural human-like voice with excellent emotional expression.

Features

• More natural and realistic voice than default TTS
• Better emotional expression for news reading
• Supports multiple voices
• Outputs MP3 format directly

Requirements

• Python 3.8+
• requests library
• FishAudio API key (get from https://fish.audio/)

Installation

1. Get your API key from https://fish.audio/
2. Add API key to your TOOLS.md or environment variable FISH_AUDIO_API_KEY
3. pip install requests (usually already installed)

Usage

Basic Text to Speech

python C:\path\to\skills\fishaudio-tts\fish_tts.py ^
--text "Your text here" ^
--output "output.mp3" ^
--voice "female"

Available Voices

• female - Default female voice (good for news)
• male - Male voice
• neutral - Neutral voice

Command Line Example

python ~/.openclaw/workspace/skills/fishaudio-tts/fish_tts.py ^
--text "各位听众好，这里是2026年3月11日新闻简报。" ^
--output "C:\Users\hyzu\Documents\openclaw\news_audio.mp3" ^
--voice "female"

Notes

• API key is required, get one for free at https://fish.audio/
• Free tier has rate limits, check FishAudio website for details
• For news video generation, female voice is recommended for better listening experience
• All generated audio files are saved locally on your machine