eze-skills plugin
eze-skills gives Claude Code two skills: a daily digest from your own sources and recording transcription with speaker labels
- 74 ★ on the repository
- Last commit
Extended access
Why this level
- Scripts fetch sources from the network and write a database and reports to the workspace, and audio-transcribe sends the recording to cloud recognition
- Alibaba Cloud and Hugging Face keys sit in `~/.claude/.env`
Install
- Works in
Manual install
git clone https://github.com/eze-is/eze-skills.gitThen cp -R eze-skills/daily-news ~/.claude/skills/ for the digest and the same for audio-transcribe.
Access and checks
How we reviewExtended accessThe tool runs code or makes network calls and only reads external data.
- Scripts fetch sources from the network and write a database and reports to the workspace, and audio-transcribe sends the recording to cloud recognition
- Alibaba Cloud and Hugging Face keys sit in `~/.claude/.env`
This is third-party code. Review the repository files before installing.
What it does
The repository publishes the author's plugin marketplace with two skills inside. Daily News runs a three-stage process: it fetches metadata from sources, writes short summaries and assembles a daily report into `output/YYYY-MM-DD.md`. The workspace keeps an interest profile in `profile.yaml`, settings in `settings.yaml`, source fetch methods in `methods/` and a SQLite database `data/news.db`. There is an RSS method in Python, and the agent adds a new source by analyzing a page and generating a method file. Duplicates are cut at three levels: by date, by a pre-check before insert and by a UNIQUE constraint in the database. Audio Transcribe turns a recording into structured text with timestamps and speaker labels. The skill normalizes audio to 16 kHz mono MP3 with ffmpeg, sends it to Alibaba Cloud DashScope for recognition, optionally separates voices with pyannote.audio and produces `.transcript.md` and `.transcript.json`. The keys `DASHSCOPE_API_KEY` and `HF_TOKEN` live in `~/.claude/.env`. The web-access skill from the same series moved to its own repository eze-is/web-access, and vc-report to eze-is/vc-report: only stubs for them remain here.
Who it is for. People who want a daily digest from their own sources or structured transcripts of meetings and interviews in Claude Code.
Good fit when
- Every morning you need a report across a dozen RSS feeds and sites filtered by your interest profile
- You need to add a new news source: the agent analyzes the page and creates a method file
- You have a meeting recording in m4a or wav and need text with timestamps and speaker labels
- A journalist or researcher needs JSON with the transcript for further processing
Not a fit when
- You need web-access or vc-report: they moved to separate repositories and only stubs remain here
- Transcription must stay fully local: audio goes to Alibaba Cloud DashScope
- You need a live news feed in real time, and a daily report is too slow
Example request
Set up daily-news in ~/daily-news, add the Hacker News RSS feed and build yesterday's digestLimitations
- Transcription needs ffmpeg, the dashscope package and a `DASHSCOPE_API_KEY` from Alibaba Cloud, and the audio goes to their recognition service.
- Speaker separation needs pyannote.audio, an `HF_TOKEN` for the first model download and acceptance of the model license.
- Documentation and skills are written in Chinese.
- The web-access and vc-report versions in this repository are not updated; current ones live in the author's separate repositories.
How to disable. For Claude Code remove the plugin with claude plugin uninstall, or the daily-news and audio-transcribe folders from `~/.claude/skills/`. The digest workspace with `news.db` and the keys in `~/.claude/.env` stay on disk and are deleted by hand.
FAQ
How do I install daily-news?
Clone the repository and copy the folder with `cp -R eze-skills/daily-news ~/.claude/skills/`. On first run the skill asks for a workspace folder and records it in `~/.claude/CLAUDE.md`.
Which sources does daily-news use?
Those you describe in `methods/`. An RSS method ships with it, and for other sites the agent creates a method file by analyzing the page and tracks the last fetch date per source.
What does audio-transcribe need?
It needs ffmpeg, the dashscope package and a `DASHSCOPE_API_KEY`. For speaker separation install pyannote.audio and provide an `HF_TOKEN` on the first model download.
Where does the audio go during transcription?
To the Alibaba Cloud DashScope speech recognition service. The skill first converts the recording to 16 kHz mono MP3 in the temporary folder `/tmp/audio-transcribe`.
Where are web-access and vc-report?
In the separate repositories eze-is/web-access and eze-is/vc-report. The web-access folder here is marked unmaintained and holds only a link.
Discussion
No questions or reviews yet.
Related
PluginsA marketplace of 18 Claude Code plugins: reply translation, a web scraper, design systems, a session log and more.
About a hundred skills for Claude Code: meetings, research, publishing, TDD, Telegram, email and personal analytics
A skill that builds an anonymized developer profile and a vertical poster from local Claude Code and other CLI history
Eight Claude Code skills for Obsidian: Granola notes, NotebookLM imports, TaskNotes tasks, recall of past sessions, chat export
daymade Skills Marketplace
Claude Code Skills Marketplace
A plugin marketplace for Claude Code and Codex: documents, audio, GitHub, session history, finance, macOS and a hardened skill-creator
Claude Skills (Alireza Rezvani)
Claude Code Skills & Plugins
A large library of skills, agents and commands by domain: engineering, marketing, product, compliance, C-level advisors and research
README in short
The repository page presents the author's collection of Claude Code skills and shows three install ways: two skills have moved to separate repositories, and daily-news is copied from this one. A skills table gives the purpose and trigger phrases. A detailed section covers daily-news: three stages and the workspace layout. Links to web-access and vc-report lead to standalone repositories.