journal-kernel

A second-brain template repo built on the agentic kernel. Use this as a starting point for your own AI-assisted journaling and knowledge management setup.

What it is

This repo provides a pre-configured agentic kernel for personal and work documentation: daily work logs, configurable personal logs, initiative tracking, planning structures, end-of-period summaries, and long-form writing support. Agents handle the repetitive parts so you can focus on capturing what matters.

It is not a notes app. It is a structured, agent-assisted writing environment that lives in a git repo.

Agents

Agent	Does what
work	Creates daily work logs, carries over tasks, writes summaries, scaffolds initiatives
personal	Creates personal logs on a configurable daily or monthly cadence, carries over tasks, writes summaries, sets up new years
scribe	Drafts, reshapes, and polishes long-form prose from notes or source material
kernel	Answers questions about the system, governs changes to it
agent-foundry	Designs and scaffolds new agents when you want to extend the system

Workflows

The personal agent defaults to monthly. Set settings.cadence on the personal agent entry in .agentic/manifest.yml to daily if you want personal logs to use YYYYMMDD.md files instead.

Workflow	Invocation	Does what
onboarding	/onboarding	First-time setup for a new journal-kernel repo
consolidate-work-month	/consolidate-work-month	Moves daily work logs into a monthly subdirectory
em-dash-check	/em-dash	Cleans Unicode em dash and en dash characters from prose
riff	/riff	Switches into discussion-first ideation mode until you ask to capture something
should-check	/should	Rewrites vague should language into explicit decision language

Getting started

Requires: akctl. Install via Go:

go install github.com/elliottpolk/akctl/cmd/akctl@latest

Then initialize your repo:

akctl init --kernel.source github.com/elliottpolk/journal-kernel

Once initialized:

1. Fill in the project: block in .agentic/manifest.yml with your own details.
2. Create .agentic/memories/state/project.md to give agents context about you and your setup.
3. Bridge the kernel to your platform:
   • GitHub Copilot: use the copilot-bridge skill
   • Claude: use the claude-bridge skill
4. Start a log.

Platform support

Agents and workflows can be bridged to GitHub Copilot (.github/) and Claude (.claude/) via the bridge skills. All canonical definitions live in .agentic/. The platform files are thin wrappers that reference them.

Built on

elliottpolk/agentic-kernel: the platform-agnostic foundation this repo extends.