
Fireworks Radio
https://github.com/yizhiyanhua-ai/fireworks-radio- Category
- AI Agents
- Rank
- No. 2111Tools index
Previous survey · No. 2106 ·
- Pricing
- Open Source
- Type
- TOOL
- Builder
- yizhiyanhua-ai
- GitHub
- 1 stars
- Date
About
Coding-session audio skill for Codex — yt-dlp + mpv fallback with stream/cache modes and lightweight preference memory.
What it does
Fireworks Radio restores terminal audio when a preferred music client cannot play. It turns search phrases into short playlists, plays public sources directly, or downloads audio before playback. Presets cover pop and ambient sessions, while a small local preference store can choose among them and prevent repeated manual selection.
Why it's ranked here
This is a focused recovery tool with unusually explicit operational boundaries. It handles failed authentication, unavailable sources, unstable networks, and overlapping players without pretending to replace a streaming service. Its narrow scope is persuasive, though broad process termination and noncommercial licensing make it unsuitable for some environments.
What's good
The two playback modes map to real conditions: direct streaming starts quickly, while local caching supports unreliable connections and repeated listening. Source switching removes both the recorded player and residual audio-only players. Preference memory stays understandable, using likes, dislikes, tags, and play counts to select from three fixed presets.
Tradeoffs
It requires macOS or Linux, network access, YouTube reachability, yt-dlp, and mpv. Playlist construction accepts the first search result, so the selected source depends heavily on query wording. Cleanup matches every audio-only mpv process, which may stop unrelated sessions. Cache mode downloads media, but the software grants no rights to that content. Commercial use needs separate permission.
How to use it well
Use it when coding and terminal media clients fail, or when you want a small repeatable focus mix without opening a browser. Start with a short playlist, choose streaming on stable networks, and cache only when permitted and useful. It does not provide a music catalog, sophisticated recommendations, media rights, or a full streaming-service client.
Technical notes+
SKILL.md defines the Codex workflow and source policy. scripts/build_playlist.sh resolves one YouTube result per query into an M3U playlist. scripts/play_mix.sh builds three hard-coded presets, supports stream and cache modes, launches mpv under nohup, stores its PID under /private/tmp, and uses pkill patterns to remove residual audio-only players. scripts/stop_mix.sh repeats that cleanup. scripts/radio_memory.py persists JSON scores and play counts, then selects the highest-scoring fixed preset. README.md and README.zh-CN.md document the same workflow, while LEGAL.md separates software permission from third-party media rights.
Observed
- License
- PolyForm Noncommercial 1.0.0 source-available license; commercial use requires separate written permission.
- Primary implementation
- Bash shell scripts, with a Python 3 helper for preference memory.
- Interface
- Codex skill plus command-line scripts for playlist creation, playback, stopping, presets, and preference feedback.
- Platform support
- macOS and Linux.
- Runtime requirements
- yt-dlp, mpv, Python 3 for preference memory, and network access to YouTube.
- Playback modes
- Direct streaming and local audio caching.
- Repository structure
- No test directory appears in the supplied repository tree.
Read from README.md, scripts/radio_memory.py, LICENSE, LEGAL.md, SKILL.md, README.zh-CN.md, scripts/play_mix.sh, scripts/stop_mix.sh, scripts/build_playlist.sh.
What it can do
Download audio from online sources
URL or source identifier → Audio file
Play audio during coding sessions
Audio file or stream → Audio playback
Stream audio without downloading
Audio stream URL → Real-time audio playback
Cache audio for offline playback
Audio stream or file → Cached audio file
Store user audio preferences
User preference settings → Saved preference data
Retrieve stored audio preferences
User request for preferences → Previously saved preference settings
Tags
Tech Stack
Comments (0)
No comments yet
Editorially curated, with community endorsements as a secondary signal. Corrections welcome.