agent-log

a-log

a-log (agent log) is a command-line journal built for AI agents. It is a fork of jrnl, extended with semantic search and designed to give agents a persistent, searchable memory.

Features

• Semantic search — find entries by meaning, not just keywords
• Plain-text, human-readable journal files
• Timestamp-aware entry parsing
• AES encryption support
• External editor integration

Quick Start

Create a new entry:

alog yesterday: Refactored the authentication module. Moved token validation into middleware.

Search by meaning:

alog --semantic "auth changes"

Everything before the first sentence-ending punctuation (., ?, !) is the title; the rest is the body. In your journal file, the result looks like:

[2025-06-15 09:00] Refactored the authentication module.
Moved token validation into middleware.

If you just call alog with no arguments, you will be prompted to compose your entry.

Origin

a-log is a fork of jrnl by Jonathan Wren, Micah Ellison, and many contributors. It retains full compatibility with jrnl's journal format while adding agent-oriented capabilities like semantic indexing and search.