work-start

name	work-start
description	Setup productivity system for work

Work Start

Initialize tasks, memory, and the dashboard.

Load memory-management and task-management before continuing. Follow the architecture, conventions, and reference templates defined by memory-management when bootstrapping or repairing the memory system.

Setup Flow

If dashboard.html is missing, copy it from ./skills/dashboard.html.

Do not use open or xdg-open. Tell the user:

• Dashboard is ready at dashboard.html. Open it from your file browser to get started.

Initialization Paths

Check whether the system is already initialized, partially initialized, or empty:

• initialized:

  • TASKS.md exists
  • CLAUDE.md exists
  • memory wiki core exists:
    • memory/SCHEMA.md
    • memory/index.md
    • memory/log.md
    • memory/glossary.md
  • memory directories exist:
    • memory/people/
    • memory/projects/
    • memory/context/
    • memory/topics/
    • memory/comparisons/
    • memory/queries/

• partial:

  • some of the above exist, but not all

• empty:

  • CLAUDE.md and the memory wiki do not exist yet

If initialized, stop after orientation:

Dashboard open. Your tasks and memory are both loaded.
- /work-update to sync tasks and refresh memory

If partial, repair the system.

If empty, run first-time bootstrap.

If TASKS.md is missing, create it using the standard task template.

Bootstrap

Only do this when CLAUDE.md and the memory wiki do not exist yet.

Questions

Ask where the user's task list lives:

Where do you keep your todos or task list?
- A local file
- An app like Asana, Linear, Jira, Notion, or Todoist
- A notes file

I'll use your tasks to learn your workplace shorthand.

Ask what the wiki should cover:

What should this wiki cover?
- My work context and operating memory
- A specific project or domain
- Something else

Build From Tasks

From the task list:

• identify names, acronyms, project references, recurring channels, and internal terms
• ask only about terms you cannot decode already
• capture aliases and nicknames carefully
• identify which items belong in hot memory versus durable wiki pages

Build a small initial context from the selected task source:

• extract high-confidence people, terms, projects, channels, and recurring context
• populate CLAUDE.md with only the most frequent operational items
• populate memory/glossary.md with decoded durable terms
• create starter people, project, and context pages only when clearly supported
• leave deeper enrichment for later updates

Example:

Task: "Send PSR to Todd re: Phoenix blockers"

I want to confirm:
1. PSR - what does it stand for?
2. Todd - who is Todd?
3. Phoenix - is this a project codename?

Build Steps

Create the system in this order:

1. TASKS.md
2. CLAUDE.md
3. memory/SCHEMA.md
4. memory/index.md
5. memory/log.md
6. memory/glossary.md
7. memory/people/
8. memory/projects/
9. memory/context/
10. memory/topics/
11. memory/comparisons/
12. memory/queries/
13. seed initial hot-cache and durable wiki content from the selected task source

Rules

• Keep CLAUDE.md lean. It is the hot cache, not the durable wiki.
• Put durable detail in memory/.
• Ask for the wiki domain before writing memory/SCHEMA.md.
• Write that confirmed domain into memory/SCHEMA.md.
• Use the memory-management reference templates for SCHEMA.md, index.md, and log.md.
• Create only the initial people, project, context, and topic pages supported by the user's current task context.
• Prefer a small high-confidence starting memory over a broad first-pass sync.
• Add every created durable page to memory/index.md.
• Record the bootstrap in memory/log.md.

End State

By the end of bootstrap:

• TASKS.md exists
• CLAUDE.md exists with a lean hot cache
• the memory wiki core and directories exist
• the glossary is seeded with first-pass decoded terms
• starter durable pages exist only where clearly supported by the selected task source
• memory/log.md uses the single-line format defined by memory-management

Repair

If the memory system exists but is incomplete:

1. Read the current memory files first.
2. Compare the existing layout against the structure defined by memory-management.
3. Create only the missing core files or directories.
4. Preserve existing content unless it clearly conflicts with the current schema.
5. Repair memory/index.md and memory/log.md if they are missing or outdated.
6. Do not rebuild the wiki from scratch unless the user asks.

Report

Productivity system ready:
- Tasks: TASKS.md (X items)
- Memory hot cache: CLAUDE.md
- Wiki core: memory/SCHEMA.md, memory/index.md, memory/log.md
- Seeded context: X people, X terms, X projects
- Dashboard: open in browser

Use /work-update to keep things current.

Notes

• If memory is partially initialized, repair it to match the current layout.
• Skip unavailable sources and note the gap.
• Memory grows organically after bootstrap.
• Prefer a small correct starting wiki over an overbuilt one.