Local Research Workflow
A privacy-first workflow for processing documents, videos, podcasts, and RSS feeds with local AI. Designed for a Mac with Apple Silicon; no cloud storage for your research data.
Estimated installation time: 60–120 minutes Requirements: macOS Sequoia or later, internet connection for downloads, an Anthropic account (for Claude Code)
The 3-phase model
Every source — paper, podcast, video, RSS article — passes through three explicit phases:
| Phase | Goal | How |
|---|---|---|
| 1 — Cast wide | Capture everything relevant | Items flow into Zotero _inbox from three sources: (1) the feedreader (feedreader-score.py) scores RSS/YouTube/podcast feeds daily and produces a sorted HTML reader and Atom feed; (2) items shared directly via the iOS share sheet; (3) manual additions from desktop/email/notes |
| 2 — Filter | You decide what enters the vault | index-score.py ranks inbox items by semantic similarity to your library; Qwen3.5:9b (local) generates a summary for mid-range items; you give a Go or No-go |
| 3 — Process | Full processing of approved items | The local subagent process_item.py fetches the full text, generates a structured literature note via Qwen3.5:9b, and writes it to the Obsidian vault — including key findings, methodology notes, relevant quotes, and [[internal links]] |
The separation between phases 1 and 3 keeps both your feed reader and your vault clean: only sources you have consciously approved end up in the vault.
Tools required
| Tool | Role | Local / Cloud |
|---|---|---|
| Zotero | Reference manager and central inbox | Local |
| Zotero MCP | Connects Claude Code to your Zotero library via local API | Local |
| Obsidian | Markdown-based note-taking and knowledge base | Local |
| Ollama | Local language model for offline tasks | Local |
| yt-dlp | Download YouTube transcripts and podcast audio | Local |
| whisper.cpp | Local speech-to-text transcription for podcasts | Local |
| NetNewsWire | RSS reader for academic and non-academic feeds | Local |
| Claude Code | AI assistant that orchestrates the workflow; generative work runs locally via Qwen3.5:9b (Ollama) | Local (default) / Cloud API with --hd |
In standard mode, only orchestration instructions are sent to the Anthropic API; all generative work is handled locally by Qwen3.5:9b. Only when --hd is explicitly requested do the prompt and source content go to the Anthropic API (Claude Sonnet 4.6). Reference data, notes, and transcriptions always stay local.
Overview of steps
- Install Homebrew (package manager)
- Install and configure Zotero 7 (including
_inboxcollection) - Set up Python environment
- Install and configure Zotero MCP
- Install Claude Code
- Install Ollama (local language model)
- Install Obsidian and create vault
- Connect everything: configure Claude Code with MCP
- Run first test
- Optional extensions (yt-dlp, semantic search, automatic updates)
- Podcast integration (whisper.cpp)
- RSS integration + feedreader filtering (NetNewsWire + feedreader-score.py)
- Spaced repetition (Obsidian plugin)
- Set up filter layer per source