Chew It

approved

by Jakob He

Analyze the current note with an LLM in the sidebar — outline, key-concept explanations, and a distilled summary. Works with Claude (Anthropic) and any OpenAI-compatible API. - This plugin has not been manually reviewed by Obsidian staff.

173 downloadsUpdated 26d agoMIT

Chew It

中文说明见 README.zh.md

An Obsidian plugin that analyzes the note you're reading with an LLM and shows the results in a side panel — without touching the original note. Open a dense document, pick the perspectives you want, click once, and get an outline, plain-language explanations of the hard concepts, and a distilled summary.

Works with Claude (Anthropic) and any OpenAI-compatible endpoint (including local Ollama and third-party proxies). You bring your own API key.

Features

  • Sidebar panel that streams the analysis token by token as Markdown. The original note is never modified.
  • Perspectives shown as checkboxes — select which ones to run, then click Chew It. Each runs in parallel into its own browser-style result tab. Three are built in (all editable):
    • Outline — hierarchy and main arguments
    • Key concepts — finds difficult ideas and explains them simply
    • Distill — strips redundancy, keeps the core points and conclusions
  • Per-perspective role — each perspective has its own optional system prompt (leave it blank to send no role).
  • Results follow the document — analysis is saved per note and restored when you reopen the app or switch back to a note. A target button jumps back to the analyzed note.
  • Per-tab actions via the ··· menu — regenerate just the active tab, or export just its result to a file.
  • Export the active tab's result as a Markdown note or a Canvas (split into cards by heading); appends to an existing file if it already exists.

Privacy / network use

Chew It sends the full text of the current note (including unsaved edits) to the LLM provider you configure, over HTTPS, using the API key you provide. Nothing is sent anywhere else. There is no length cap: a note longer than the model's context window is automatically split into parts, analyzed in sequential calls, and the outputs are stitched into one seamless result — mind billing on very long notes, since they cost several calls.

Installation

From the Community Plugins browser (once approved)

Settings → Community plugins → Browse → search Chew It → Install → Enable.

Manual

  1. Download main.js, manifest.json, and styles.css from the latest release.
  2. Copy them into <vault>/.obsidian/plugins/chew-it/.
  3. Reload Obsidian, then enable Chew It under Settings → Community plugins.

From source

npm install
npm run build      # produces main.js

Copy main.js, manifest.json, and styles.css into your vault's .obsidian/plugins/chew-it/ folder. Use npm run dev to watch and rebuild while developing.

Usage

  1. In Settings → Chew It, choose a provider and enter your API key (for OpenAI-compatible endpoints, also set the base URL and model name).
  2. Open a Markdown note.
  3. Click the ✨ ribbon icon, or run the command Analyze current note. The panel opens on the right.
  4. Tick the perspectives you want and click Chew It. Switch result tabs to read each one.
  5. Use the ··· menu on a tab to regenerate it or export its result.

Configuration

Everything is editable in Settings → Chew It:

  • Provider — Claude (Anthropic), default model claude-opus-4-8; or an OpenAI-compatible endpoint with a custom base URL and model.
  • Perspectives — add, rename, enable/disable, and edit each one's role and prompt.
  • Output templates — configure export targets (Markdown / Canvas), destination folder, and filename template (placeholders: {note} {label} {date} {datetime}).

The interface language (English / 中文) follows your Obsidian display language automatically.

Token limits are handled automatically — long notes are split into multiple calls and stitched back together; there is nothing to configure.

Support

If Chew It saves you time, you can buy me a coffee on Ko-fi

License

MIT

For plugin developers

Search results and similarity scores are powered by semantic analysis of your plugin's README. If your plugin isn't appearing for searches you'd expect, try updating your README to clearly describe your plugin's purpose, features, and use cases.