inbox | log | projects | areas | resources | archive | search

journal

A personal work journal combining PARA (Projects, Areas, Resources, Archive) with Bullet Journal conventions, designed to be used by both humans and AI coding agents.

Why

Context gets lost between sessions. Whether it's a conversation with Claude, a Gemini session, or just switching between projects during the day, the "what was I doing and why" fades fast. This repo is the single source of truth for that context — a place where every AI tool and every work session can read from and write to.

It blends two proven productivity systems:

• PARA organizes what you're working on into four categories by actionability
• Bullet Journal provides the rhythm — daily logs, quick capture, periodic review, and migration

The result is a plain-text system that works from the terminal, renders on GitHub, and is readable by any AI tool.

Structure

journal/
├── inbox.md                          # Quick capture — unsorted tasks and notes
├── workqueue.md                      # Legacy `/next` queue (superseded by ~/.whisper/)
├── projects/                         # Active work with defined goals
│   └── _index.md
├── areas/                            # Ongoing responsibilities (no end date)
│   ├── _index.md
│   └── log/                          # Daily session logs
│       └── YYYY/YYYY-MM/YYYY-MM-DD.md
├── resources/                        # Reference material and notes
│   ├── _index.md
│   └── skills/                       # AI agent skills (slash commands)
│       ├── morning/    capture/    jlog/
│       ├── renew/      park/       next/
│       ├── close/      wrap-up/
│       ├── migrate/    standup/    weekly/
│       └── archive/
├── archive/                          # Completed projects, inactive areas
│   ├── _index.md
│   ├── queue/                        # Archived pre-whisper `/next` snapshots
│   └── whisper/                      # Knowledge store (~/.whisper → symlink here)
├── bin/
│   ├── build                         # Static site build (pandoc)
│   ├── doctor                        # Self-test of the journal chain
│   ├── generate-indexes              # Site: directory listings
│   ├── generate-search-index         # Site: lunr search index
│   ├── convert-org                   # Org → markdown helper
│   └── sync                          # (deprecated stub) skills install via npm:incitaciones
├── AGENTS.md                         # Instructions for AI tools
├── CLAUDE.md -> AGENTS.md            # Symlink for Claude Code
├── .mise.toml                        # Tool versions + task runner
└── .zk/                              # Zettelkasten config for org-mode notes

The top level is pure PARA: projects/, areas/, resources/, archive/, plus inbox.md as the entry point. Everything else nests inside those four categories — daily logs are an ongoing area (areas/log/), skills are a resource, and session snapshots/knowledge live in archive/whisper/ (which ~/.whisper symlinks to). The old queue/ directory of /next snapshots is superseded by whisper notes and kept at archive/queue/.

PARA in brief

Items flow naturally: an inbox capture becomes a project task, a completed project moves to archive, a useful finding becomes a resource.

Bullet notation

Tasks in markdown files use extended checkbox syntax:

The key BuJo practice is migration: periodically reviewing open items and deciding whether to keep, move, schedule, or cancel them. The trail of [>] markers tells you where things went.

Daily log

Each day gets a file at areas/log/YYYY/YYYY-MM/YYYY-MM-DD.md:

# 2026-04-23

## Focus
- Refresh journal docs for the latest session workflow

## Log

### Sessions 17:40 (journal — 2 sessions)
- Updated README.md to document the new `/next` and `/wrap-up` skills
- Added `workqueue.md` to the documented repo structure and clarified that it is agent-managed
- **Arc:** moved the docs from a direct-log-only model to a direct-log + queued-handoff model
- **Next:** keep examples in sync as skills evolve

## Tasks created

## Decisions
- `/park` writes directly to today's log
- `/next` queues work for later synthesis by `/close` or `/wrap-up`

Sessions are appended throughout the day. /park writes a quick snapshot directly to today's log, /next stashes a fast handoff into ~/.whisper/, /close wraps up the current slug and folds in any whisper notes queued for it, and /wrap-up processes any remaining queued sessions across projects.

AI agent skills

Shared session-lifecycle skills (close, next, park, renew) come from incitaciones (npm), managed by its installer into ~/.agents/skills/. Journal-specific skills live in resources/skills/, symlinked into ~/.agents/skills/. bin/sync is a deprecated stub.

How the handoff commands relate

• Use /park when you're switching context and want an immediate note in today's log without committing the journal.
• Use /next when you need a fast checkpoint; it snapshots an agent-managed note to ~/.whisper/ for later synthesis.
• Use /close when you're done with the current slug; it can fold in queued /next entries for that same slug, even from earlier dates.
• Use /wrap-up at the end of the day to process any remaining queued sessions across all slugs.

Whisper notes are normally agent-managed state. Edit them manually only for recovery or cleanup.

Typical flow

/morning            → see what's open, set focus
  ... work ...
/capture fix the CI → add to inbox
  ... work ...
/next               → stash a quick handoff in ~/.whisper when you need to pause fast
/renew              → infer and load context for the next thing
/renew other-proj   → explicitly load context for a project/area
  ... work ...
/close              → log this session, folding in any queued notes for the same slug
/wrap-up            → end-of-day sweep across all queued sessions

/close can fold in queued /next entries for the current slug across dates, while /wrap-up processes whatever remains in the queue across all slugs.

Installing skills

Shared skills install with the incitaciones installer; journal skills are symlinked from resources/skills/ — see resources/skills/README.md.

Tasks (via mise)

All tasks are defined in .mise.toml and run with mise run <task>:

Setup

Environment variable

Set $JORNAL to the repo path so skills can find it from any directory:

# Fish
set -gx JORNAL $HOME/para/areas/dev/gh/ak/journal

# Nushell
$env.JORNAL = $"($env.HOME)/para/areas/dev/gh/ak/journal"

# Bash/Zsh
export JORNAL="$HOME/para/areas/dev/gh/ak/journal"

If unset, skills default to ~/para/areas/dev/gh/ak/journal.

Install tools and deploy

cd $JORNAL
mise install          # installs pandoc, zk
mise run whisper-sync # commit + push new whisper entries

Dependencies

Managed by mise (.mise.toml):

• pandoc — document conversion (org→md, md→html)
• zk — note management (optional, for org-mode journal)

No dependencies are required for the AI skills — they're plain markdown files.

License

CC0 1.0 Universal — public domain.