Daily Digest for Obsidian
“Your day, distilled.”
What survives a workday: commits, a merged PR, a closed ticket. What doesn’t: why I threw out the first approach. The searches that shifted my thinking. The rabbit hole that cost two hours and saved six.
I built this because I kept losing that context. After the third time I re-solved the same problem from scratch, I figured I needed a record. Daily Digest captures all of it and turns it into something you can actually read tomorrow — or six months from now.
Everything is off by default. You choose which data sources to enable and whether you want AI summaries. With a local model (Ollama, LM Studio), nothing ever leaves your machine.
What it does
Browser history, searches, Claude Code sessions, Codex CLI sessions, and git commits — everything in one pipeline.
Scrubs API keys, tokens, passwords, and sensitive URLs before anything reaches your vault or the cloud.
AI-generated headline, key themes, notable moments, and reflection questions tailored to your day.
A structured Markdown note with Dataview-ready frontmatter and a Notes section that's always preserved across regenerations.
Data sources
| Source | What it reads | How |
|---|---|---|
| Browser history | URLs, page titles, timestamps | SQLite read-only copy |
| Search queries | Queries from Google, Bing, DuckDuckGo, Kagi, Perplexity | Extracted from browser history URLs |
| Claude Code sessions | Your prompts to Claude Code (not responses) | ~/.claude/projects/**/*.jsonl |
| Codex CLI sessions | Your prompts to Codex CLI (not responses) | ~/.codex/history/*.jsonl |
| Git commits | Commit messages, timestamps, file change stats | Local .git directories under a configurable parent folder |
Screenshots
Privacy architecture
When Anthropic’s cloud API is used, a 4-tier escalation chain controls exactly how much data leaves your machine. The plugin always selects the most private tier available given your configuration.
Aggregated statistics only. No per-event data — just topic distributions, focus scores, and temporal shapes.
Structured abstractions only — activity types, topics, entities. No raw URLs or queries sent.
Budget-proportional summaries — domain counts, top titles, queries. No full URLs.
Full context (sanitized). All data types included. Used only with local models by default.
Get Daily Digest
Download from the latest release and drop it into your Obsidian vault's plugin folder.