mirror-log An append-only event log for capturing thoughts, notes, and data you don't want to lose.

Philosophy This is a lab notebook for your brain. Write things down once, never lose them, find them later. No editing history, no second-guessing what you wrote, no normalization that loses context.

Think of it as a foundation for personal knowledge that:

Accepts anything (text, documents, conversation exports) Never modifies what you wrote Lets you search and connect ideas later Stays local, debuggable, and honest Core Principles SQLite is the source of truth - Your data, your filesystem, no servers Append-only - Events are never modified or deleted No magic - Every query is visible SQL, every function does one thing Source tracking - Know where everything came from Debuggable - Open mirror.db in sqlite3 anytime and understand it

Installation

bash git clone https://github.com/CromboJambo/mirror-log cd mirror-log cargo build --release

The binary will be at target/release/mirror-log.

Optionally, add it to your PATH:

bash cargo install --path .

Quick Start

Add a thought

mirror-log add "Overhead allocation needs review" --source journal

Add from a file

mirror-log add "$(cat meeting-notes.md)" --source meetings

Bulk import (one line per event)

cat ideas.txt | mirror-log stdin --source ideas

View recent entries

mirror-log show --last 10

Search

mirror-log search "overhead"

Filter by source

mirror-log show --last 20 --source journal

Database info

mirror-log info Usage Adding Events Single entry:

bash mirror-log add "Your content here" --source cli With metadata:

bash mirror-log add "Meeting notes"
--source meetings
--meta '{"attendees": ["alice", "bob"], "project": "Q4-planning"}' From file:

bash mirror-log add "$(cat document.md)" --source documents Bulk import from stdin:

bash cat file.txt | mirror-log stdin --source import seq 1 100000 | mirror-log stdin --source test # Fast bulk import Reading Events Recent entries:

bash mirror-log show --last 20 Filter by source:

bash mirror-log show --last 50 --source journal Search content:

bash mirror-log search "cost allocation" Get specific event:

bash mirror-log get Database statistics:

bash mirror-log info Direct SQLite Access The database is just SQLite. You can query it directly:

bash sqlite3 mirror.db

See recent events

SELECT datetime(timestamp, 'unixepoch'), source, content FROM events ORDER BY timestamp DESC LIMIT 10;

Count by source

SELECT source, COUNT(*) FROM events GROUP BY source;

Search with SQL

SELECT * FROM events WHERE content LIKE '%overhead%' ORDER BY timestamp DESC; Schema sql CREATE TABLE events ( id TEXT PRIMARY KEY, -- UUID timestamp INTEGER NOT NULL, -- Unix epoch source TEXT NOT NULL, -- Where this came from content TEXT NOT NULL, -- The actual data meta TEXT -- Optional JSON metadata ); Simple. Visible. No surprises.

Data Model Every event has:

id: UUID (unique identifier) timestamp: Unix epoch (when it was logged) source: String tag (cli, journal, meetings, import, etc.) content: The actual text/data meta: Optional JSON for structured metadata Events are never modified or deleted. The log is append-only.

Use Cases Personal journal:

bash mirror-log add "$(date): Thought about variance analysis today" --source journal Meeting notes:

bash mirror-log add "$(cat meeting-2024-01-13.md)" --source meetings Conversation exports:

bash mirror-log add "$(cat 'claude-chat.md')" --source claude-chat Git history:

bash git log --oneline | mirror-log stdin --source git-history Research clips:

bash mirror-log add "Interesting paper on ABC costing..."
--source research
--meta '{"url": "https://...", "tags": ["costing", "ABC"]}' Performance Bulk imports use transactions for speed:

10,000 events: ~1 second 100,000 events: ~3 seconds 1,000,000 events: ~30 seconds SQLite handles millions of events efficiently. Text is tiny.

What's NOT Here (By Design) ❌ No updates or deletes ❌ No migrations (schema is stable) ❌ No AI/embeddings (yet - that's a separate layer) ❌ No web UI (CLI first, UI later) ❌ No sync (local-first) ❌ No normalization (events are atomic) These features can be built on top without touching the core log.

Future Layers This is the foundation. Future additions will be separate systems that read from events:

Annotations table - Your thoughts about events (interpretive layer) Relations table - Connections between events (manual or AI-suggested) Embeddings - Vector search for semantic similarity Views - Saved queries and perspectives Web UI - Read-only interface for browsing Export tools - Markdown, JSON, etc. All of these will reference events by ID but never modify them.

Philosophy: The Lab Notebook This is inspired by the lab notebook principle:

Write in pen, never erase Date everything Note the source Preserve context Never lose data Digital tools make it too easy to edit, delete, and lose context. This tool enforces honesty by making the log immutable.

Later analysis, connections, and interpretations happen in separate layers. The original data stays pristine.

License AGPL-3.0-or-later

This program is free software: you can redistribute it and/or modify it under the terms of the GNU Affero General Public License as published by the Free Software Foundation, either version 3 of the License, or (at your option) any later version.

This ensures that if anyone runs a modified version of this software as a network service, they must make the source code available to users of that service.

See the LICENSE file for full text.

Contributing This is a personal tool first. If it's useful to you, great. If you want to contribute, open an issue and let's talk.

Keep it simple. Keep it honest. Keep it debuggable.

AI coding assistants from: claude, chatgpt, gpt-oss, devestral etc... thank you to HuggingFace, Unsloath, OpenCode, Llamma.cpp