Documentation on Arivu

Getting Started

Wed, 08 Jul 2026 00:00:00 +0000

Arivu runs as a single Go application with an embedded browser interface and SQLite database. A model provider is optional.

Run Arivu Locally

Install Go, clone the repository, and start the server:

git clone https://github.com/glnarayanan/arivu.git
cd arivu
go run ./cmd/arivu serve --addr 127.0.0.1:8080 --db arivu.sqlite3

Open http://127.0.0.1:8080/auth, create an account, and sign in. For a production Linux VPS, use the first-party installer instead.

Capture Something

Choose Capture from any authenticated screen, then select Link, Note, Quote, or File. Saving does not require a folder, tag, workflow stage, or AI configuration.

Self-Hosting Arivu

Wed, 08 Jul 2026 00:00:00 +0000

Arivu’s core runs as one Go application with an embedded browser UI, API, workers, CLI commands, and SQLite database. The first-party installer is the recommended production path and can additionally manage the isolated native runtime for complete browser capture.

One-Command Install

curl -fsSL https://install.arivu.app | sudo bash

The bootstrap downloads arivu-installer, verifies it against release checksums, installs it under /usr/local/bin, and starts interactive setup. It does not require the GitHub CLI or a GitHub login.

Environment Variables

Wed, 08 Jul 2026 00:00:00 +0000

Installer-managed deployments keep /etc/arivu/arivu.env focused on machine configuration. Routine public URL, signup, cookie, model-provider, email, and X settings can be changed by an admin at runtime.

Core Runtime

Variable	Default	Purpose
`ARIVU_ADDR`	`:8080`	HTTP listen address
`ARIVU_DB`	`arivu.sqlite3`	SQLite database path
`APP_URL`	`http://localhost:8080`	Public URL and callback base
`SECRET_KEY`	development fallback	Sessions and encrypted runtime settings
`ADMIN_EMAILS`	empty	Comma-separated administrator emails
`SIGNUPS_ENABLED`	`true`	Public account registration
`COOKIE_SECURE`	`false`	Secure browser cookies, required behind production HTTPS
`ARIVU_FETCH_USER_AGENT`	`Arivu/2.0`	Server-side bookmark fetch identity

Generate a unique production secret and keep it outside version control:

Administration & Recovery

Wed, 15 Jul 2026 00:00:00 +0000

Arivu keeps administration separate from ordinary personal knowledge work. Administrators can manage users, runtime settings, API usage, system status, audit history, and processing recovery without exposing those controls to non-admin users.

Inspect Processing Failures

Open Administration, then API Usage to see durable failed jobs and failed summaries. Entries can include the owner, bookmark, attempt count, timestamps, provider and model context, safe error codes, and recorded failure details.

The two lists answer different questions:

X Bookmarks Integration

Mon, 15 Jun 2026 00:00:00 +0000

Arivu can import bookmarks from X when an administrator enables the optional OAuth integration. Imported links enter the same Library and background processing pipeline as other captures.

Prerequisites

An X developer application with OAuth 2.0 enabled
Confidential client credentials
A public HTTPS APP_URL

Configure the callback as:

https://your-domain.example/settings?section=connections

Request the scopes bookmark.read, tweet.read, users.read, and offline.access.

Configure Arivu

Prefer Admin settings for routine runtime configuration. The environment fallback is:

X_INTEGRATION_ENABLED=true
X_CLIENT_ID=your_x_client_id
X_CLIENT_SECRET=your_x_client_secret
APP_URL=https://your-domain.example

Set X_REDIRECT_URI only when you need to override the callback derived from APP_URL.

Browser Extension Setup

Thu, 19 Feb 2026 00:00:00 +0000

The Manifest V3 extension captures pages, links, and selected text into your own Arivu instance. Custom domains are configured from the popup. You do not need to edit manifest.json or source files.

Install an Unpacked Development Copy

Chrome

Open chrome://extensions/.
Enable Developer mode.
Choose Load unpacked.
Select the repository’s extension/ directory.

Firefox

Open about:debugging#/runtime/this-firefox.
Choose Load Temporary Add-on.
Select extension/manifest.json.

Connect Your Instance

Open the extension popup and choose Settings.
Enter https://your-domain.example/api as the API URL.
Approve the browser permission request for your Arivu origin.
Sign in to Arivu in the same browser profile and visit the app once.

The extension dynamically registers the token content script for the approved origin. It exchanges your web session for an audience-scoped extension token through /api/auth/extension-token; no manifest modification is required.

Capture & Library

Sat, 11 Jul 2026 00:00:00 +0000

Capture is available from every authenticated screen. Library is where captured and derived knowledge becomes browsable without forcing everything into folders first.

Capture Without Prerequisites

Choose Capture, then select Link, Note, Quote, or File. A capture does not require a tag, folder, stage, classification, embedding, or model provider.

Link saves a page through the normal bookmark pipeline.
Note creates a standalone note.
Quote preserves selected evidence with its source URL and an optional interpretation.
File imports supported material as a searchable note.

The browser extension, PWA share target, CLI, and local automation routes use the same owned knowledge store. Link and quote captures can queue in the browser while offline.

Second Brain Workflow

Sat, 04 Jul 2026 00:00:00 +0000

Arivu is organized around one loop: Capture -> Connect -> Discover -> Learn. You can begin without choosing a taxonomy or configuring AI, then add structure as your knowledge base grows.

Capture

Use the global Capture action for a link, note, quote, or file. Browser extension, PWA share target, and CLI capture remain available. Imported documents and transcripts become searchable notes instead of entering a separate media library.

New material appears in Library. Inbox remains a filtered Library view for users who want to assign a stage, importance, or next action. One-time feed imports can seed the Library, while ongoing RSS and Atom subscriptions continue sending accepted entries through the same pipeline.

Preservation & Sharing

Mon, 13 Jul 2026 00:00:00 +0000

Arivu keeps private research evidence and deliberate public publishing as separate workflows. Direct capture remains useful without the isolated native capture runtime or a public link.

What Normal Capture Keeps

A link follows the ordinary capture pipeline. Safe direct extraction and isolated rendered capture can run together. Complete direct evidence becomes readable without waiting for the browser runtime, while stronger rendered evidence can improve a weak reader later. A failed browser run or challenge page cannot replace stronger evidence or the last good reader.

Notes, Links & Research Objects

Sat, 04 Jul 2026 00:00:00 +0000

Notes is a primary destination alongside Home, Library, Graph, and Insights. It is the writing workspace for ideas that begin with you rather than with a saved page.

Notes and Daily Notes

Create a standalone note from global Capture or the Notes page. A note has its own detail route and can carry explicit links, backlinks, tasks, reminders, and related material. Home also includes a dated daily note for lightweight working context.

Reminders & Tasks

Sat, 04 Jul 2026 00:00:00 +0000

Tasks and reminders support knowledge work without turning Arivu into a full project-management system. Attach them to a bookmark or note so the action stays beside its evidence.

Tasks

A task, also called an action item, is an undated checklist item. Create one from relevant item details, complete it when finished, or remove it when it no longer matters. Pending work can appear on Home and in its Focus or Review contexts.

AI Assistant

Sun, 05 Jul 2026 00:00:00 +0000

AI is an optional layer in Arivu, not a prerequisite for capture, organization, or retrieval. The canonical Search / Ask surface can synthesize a cited answer, while Assistant action review keeps proposed changes under your control.

Cited Ask

Ask deterministically assembles a cited response from saved content in your Arivu account and does not call a model provider. Its claims link back to owned bookmarks or notes. Optional providers can improve summaries and other enrichment stored with that material, while ordinary full-text Search and cited Ask remain available without one.

Search & Ask

Sat, 11 Jul 2026 00:00:00 +0000

Search / Ask is globally available and has the canonical route /search. It supports two related jobs: finding saved material and synthesizing a cited answer from it.

Search Your Saved Knowledge

Search retrieves owned bookmarks and notes with snippets and source links. Use text and structured filters to narrow the result set. Full-text retrieval is local and does not require a model provider or embeddings.

Semantic-similarity relationships are separate from full-text search. They appear in connected surfaces only when embeddings exist, while ordinary retrieval continues without them.

Knowledge Graph & Topic Evolution

Fri, 20 Feb 2026 00:00:00 +0000

Graph helps you inspect how owned knowledge connects. It opens a bounded recent view or a focused neighborhood instead of attempting to draw your entire collection at once.

Nodes and Relationships

Graph can project bookmarks, notes, daily notes, annotations, knowledge objects, entities, and concepts. Relationships include:

explicit links you created
source relationships
shared concepts
shared entities
semantic similarity when embeddings exist

Explicit links are canonical. Derived relationships are quieter, rebuildable, and include provenance and confidence where relevant. A missing model provider or embedding does not remove explicit, source, concept, or entity relationships.

Collections & Tags

Fri, 20 Feb 2026 00:00:00 +0000

Collections and tags remain optional ways to organize bookmarks. Arivu does not require either before capture, and Library adds broader filters across all supported knowledge types.

Collections

A collection is a named group of bookmarks. Collections can be nested into an owner-scoped tree, and a bookmark can belong to more than one collection. Deletion is non-recursive: move or delete child collections before deleting a parent. Deleting a leaf removes its memberships, never its bookmarks.

Insights

Fri, 20 Feb 2026 00:00:00 +0000

Insights turns locally detected patterns into explainable prompts for learning. The canonical route is /insights; the existing /analytics link remains compatible.

What Insights Detects

Current insight families include:

Emerging themes appearing across recent material
Recurring connections between sources or notes
Forgotten value worth revisiting
Knowledge gaps suggested by the shape of your collection
Serendipitous connections between otherwise distant items

Analytical Insights include an explanation, time window, detector confidence, detection reason, supporting items you own, and possible next actions. Evidence links let you inspect why the pattern appeared.

Import & Export

Mon, 15 Jun 2026 00:00:00 +0000

Imports enter the same Library as ordinary captures. Documents and transcripts become notes, while full JSON backups preserve durable account knowledge for restore.

Bookmark and Feed Imports

Settings accepts browser bookmark HTML, an Arivu JSON backup, one URL per line, OPML, RSS or Atom, and URL-bearing CSV or TSV files. Common Pocket, Raindrop, Linkwarden, Readwise, and Kindle-style exports work when their fields can be mapped.

Import jobs run in the background and report fetched, processed, skipped, and failed items. Arivu records source provenance, deduplicates normalized URLs, and rejects unsafe URLs before creating bookmarks.

Resurfacing & Rediscovery

Mon, 06 Jul 2026 00:00:00 +0000

Arivu brings older material back when it has a reason to matter. Rediscovery supports the Learn stage of the core loop without turning your collection into a guilt queue.

Where Material Returns

Home can show one useful memory alongside new material, active work, and recent notes.
Review remains a contextual Home view for items that are due, important, unfinished, stale, or otherwise worth another look.
Insights detects forgotten value, recurring connections, gaps, themes, and serendipitous relationships with supporting evidence.

Each review candidate or insight explains why it appeared. You can inspect the source before deciding what to do.

API Reference

Mon, 15 Jun 2026 00:00:00 +0000

The Go application serves its API under /api. This page is a route overview; the application repository at the pinned release is the source of truth for request and response fields.

Authentication Audiences

Web uses HTTP-only cookies and CSRF protection for browser mutations.
CLI uses bearer tokens from /api/auth/cli/login.
Extension uses audience-scoped tokens issued through an authenticated web session.

Tokens cannot cross audience boundaries. Every owned query and mutation is scoped to the authenticated user.

Keyboard Shortcuts

Fri, 20 Feb 2026 00:00:00 +0000

Arivu relies on familiar browser controls and a discoverable set of application shortcuts. Press ? outside a form field to open the complete in-app guide. Single-key commands stay disabled while you type in a form or use a dialog.

Global

Keys	Action
`Cmd/Ctrl + K`	Open the command palette for navigation, Capture, Search / Ask, and contextual actions
`Q`	Open Capture
`/` or `F`	Focus Search
`Cmd/Ctrl + P`	Print the current view
`?`	Open the shortcut guide
`Escape`	Close the active dialog, menu, or annotation composer and restore focus where applicable
`Tab` / `Shift + Tab`	Move through interactive controls and trapped dialog focus

Persistent Capture, Search, and More actions show compact key hints on larger screens. Mobile keeps the controls visually restrained.

Command Line Interface

Wed, 08 Jul 2026 00:00:00 +0000

The same Go binary serves the app and provides CLI commands. CLI sessions use their own bearer-token audience and cannot be reused as browser or extension sessions.

Build and Run

git clone https://github.com/glnarayanan/arivu.git
cd arivu
go build -trimpath -ldflags="-s -w" -o arivu ./cmd/arivu
./arivu version
./arivu serve -addr 127.0.0.1:8080 -db arivu.sqlite3

Open http://127.0.0.1:8080/auth and create an account before CLI login.

Log In

./arivu login \
  --api http://127.0.0.1:8080/api \
  --email you@example.com \
  --password your-password

The CLI stores its API URL and tokens in the local user configuration directory. ARIVU_API supplies a default URL when --api is omitted.