divesh.gg

Personal blog built with Astro.

• Register divesh.gg domain

Commands

Command	Action
bun install	Install dependencies
bun dev	Start dev server at localhost:4321
bun build	Build production site to ./dist/
bun preview	Preview build locally
bun sync api --export-library	Fetch API sources into data/knowledge.db, then export /library data
bun sync --export-library	Export current data/knowledge.db to /library data without fetching APIs
bun sync --stats	Show local knowledge DB stats
bun zirp inventory	Inventory the local divesh_zirp corpus
bun zirp init-db	Create namespaced zirp_* tables in data/knowledge.db
bun zirp index	Build the local SQLite/FTS oracle index
bun zirp serve	Open the local divesh_zirp web cockpit + SQLite explorer on 127.0.0.1:7331
bun zirp search "games as training grounds"	Search the personal oracle corpus
bun zirp ask "what should I write next?"	Ask the oracle; uses OpenAI/Anthropic if configured, otherwise prints the prompt

Linting & Formatting

This project uses Biome for linting and formatting.

Command	Action
bun check	Type check (Astro/TypeScript)
bun check:all	Type check + lint + format check
bun lint	Check for linting issues
bun lint:fix	Fix linting issues
bun format	Format all files

Run Biome directly on specific files with bunx biome check <file>.

Project Structure

src/
├── components/      # UI components
├── content/
│   ├── waves/       # Short posts
│   └── depths/      # Long-form content
├── layouts/         # Page templates
├── pages/           # Routes
├── scripts/         # Client-side JS
└── styles/          # Global CSS
scripts/
├── db.ts            # SQLite knowledge base helpers
├── sync.ts          # Sync/export CLI
└── zirp.ts          # Local-first personal oracle CLI
docs/
└── zirp/            # divesh_zirp architecture, persona docs, and PR notes
public/
└── data/            # Generated/static data files

Features

Books (/books)

Interactive library explorer powered by Goodreads CSV export. Filter by shelf, decade, or recommender. Sort by various fields. Export selections as JSON or text.

Music (/music)

Music explorer for Spotify listening data.

Library (/library)

Digital garden for Readwise and Zotero resources. Data flows through local SQLite, then exports to public/data/library.json for Astro.

bun sync api --export-library  # normal incremental sync + export
bun sync readwise --full       # force full Readwise backfill
bun sync zotero --full         # force full Zotero refresh
bun sync readwise --since 2026-04-01T00:00:00Z
bun sync --stats

Required .env values:

READWISE_TOKEN=...
ZOTERO_API_KEY=...
ZOTERO_USER_ID=...

Notes:

• First Readwise sync can take several minutes for large libraries.
• Later Readwise/Zotero runs are incremental via saved sync_state in data/knowledge.db.
• data/ and generated JSON exports are ignored; regenerate them locally.

Personal Oracle (divesh_zirp)

Local-first oracle inspired by vgr_zirp. It indexes blog posts, selected notes, Versa context, Readwise/Zotero metadata, Goodreads, and Spotify into zirp_* tables inside data/knowledge.db.

Canonical refresh:

bun sync api --export-library  # remote APIs → data/knowledge.db → public/data/library.json
bun zirp inventory             # dry-run: what ZIRP sees now
bun zirp index                 # write/rebuild zirp_* tables
bun zirp stats                 # inspect indexed corpus
bun zirp serve                 # local cockpit + SQLite explorer

Use it:

bun zirp search "games as training grounds"
bun zirp prompt "connect WoW, feedback loops, and Versa"
bun zirp ask "what should I write next?"

Privacy defaults: journal/ is excluded unless --include-journal is passed; API calls only happen when a model API key is configured. ask defaults to OpenAI gpt-5.5 with reasoning effort low when OPENAI_API_KEY is present, with Anthropic fallback.

See docs/zirp/DX.md for the canonical runbook.

Content

Posts are Markdown files in src/content/ with frontmatter:

---
title: string
description: string
author: string
pubDate: Date
tags: string[]
image?: { url: string, alt: string }
---