productivity-setup

name	productivity-setup
description	Initialize productivity system for a project: .productivity.yml connector config, TASKS.md, and workplace context.
argument-hint	["project path"]
disable-model-invocation	true
allowed-tools	["Read","Write","Glob","Grep"]

Productivity Setup

First-run entry point. Creates the per-project connector config, initializes task tracking, and bootstraps workplace context.

Use skills/context-advisor/SKILL.md when owner, privacy boundary, cross-repo promotion, or write authority is unclear.

Owner Routing

• Use the owning repo as the local task and context surface by default.
• If context-advisor is unavailable, read skills/context-advisor/references/storage-routing.md directly.
• Do not create new context/ folders for durable context.

Instructions

Resolve the owning repo before creating or updating local task and context files.

1. Check What Exists

Check the working directory for:

• .productivity.yml -- connector config
• TASKS.md -- task list
• CLAUDE.md or .claude/CLAUDE.md -- one canonical hot context file
• context/ -- durable context directory

2. Configure Connectors

If .productivity.yml exists: Show current config and ask if the user wants to update it.

If .productivity.yml doesn't exist: Present each connector category as a numbered choice:

Let's configure which external sources to connect for this project:

1. Calendar:
   (1) Google Calendar  (2) Microsoft 365  (3) gogcli (multi-account)  (4) None

2. Email:
   (1) Gmail  (2) Microsoft 365  (3) gogcli (multi-account)  (4) None

3. Project Tracker:
   (1) Jira  (2) Asana  (3) Linear  (4) GitHub Issues  (5) Monday  (6) ClickUp  (7) None

4. Knowledge Base:
   (1) Notion  (2) Confluence  (3) None

5. Chat:
   (1) Slack  (2) None

6. Messages:
   (1) iMessage  (2) None

7. Contacts:
   (1) gogcli  (2) None

When gog is selected for any connector:

• Prompt: "Which Google account? (e.g., [e-mail])"
• Write the connector value as gog (not gogcli)
• Write the account as <connector>-account: <email> (e.g., calendar-account: [e-mail])
• Different connectors can use different Google accounts

Write .productivity.yml with the selected values. Reference the productivity-connectors skill for the connector value names.

3. Create What's Missing

If TASKS.md doesn't exist: Create it with the fallback template from the productivity-tasks skill. Place it in the owning repo root.

If context/ doesn't exist: Create the scaffold directory structure with .keep files so the LLM and sync skill know where things go:

context/
  glossary.md          # empty; acronyms, terms, nicknames, codenames
  people/.keep         # individual profiles, one file per person
  projects/.keep       # project details, one file per project
  reference/.keep      # company, teams, tools, processes
  personal/.keep       # user-owned local context for this repo

Create glossary.md as an empty file (it's a single file, not a directory). Create .keep in each subdirectory.

If no canonical repo-level CLAUDE.md and no context/ exist: This is a fresh setup; begin the context bootstrap workflow (see Step 5).

Interpret "current working directory" as the owning repo, not merely the shell cwd.

4. Orient the User

If everything was already initialized:

Productivity system configured. Your tasks and context are ready.
- /productivity-sync to sync tasks and check context
- /productivity-sync --deep for a comprehensive scan of all activity

If context hasn't been bootstrapped yet, continue to Step 5.

5. Bootstrap Context (First Run Only)

Only do this if no canonical repo-level CLAUDE.md and no context/ exists yet.

The best source of workplace language is the user's actual task list. Real tasks = real shorthand.

Ask the user:

Where do you keep your todos or task list? This could be:
- A local file (e.g., TASKS.md, todo.txt)
- An app (e.g. Asana, Linear, Jira, Notion, Todoist)
- A notes file

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

Once you have access to the task list:

Reference the productivity-connectors skill for MCP tool names if the user points to an external app. If the tool is unavailable, ask the user to paste or describe their tasks instead.

For each task item, analyze it for potential shorthand:

• Names that might be nicknames
• Acronyms or abbreviations
• Project references or codenames
• Internal terms or jargon

For each item, decode it interactively:

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

I see some terms I want to make sure I understand:

1. **PSR** - What does this stand for?
2. **Todd** - Who is Todd? (full name, role)
3. **Phoenix** - Is this a project codename? What's it about?

Continue through each task, asking only about terms you haven't already decoded.

6. Optional Deep Scan

After task list decoding, offer:

Do you want me to do a deep scan of your messages, emails, and documents?
This takes longer but builds much richer context about the people, projects,
and terms in your work.

(1) Yes, deep scan
(2) No, stick with what we have

If they choose deep scan:

Gather data from available MCP sources (reference the productivity-connectors skill for tool names, skip unavailable sources):

• Chat: Recent messages, channels, DMs
• Email: Sent messages, recipients
• Documents: Recent docs, collaborators
• Calendar: Meetings, attendees

Build a braindump of people, projects, and terms found. Present findings grouped by confidence:

• Ready to add (high confidence) -- offer to add directly
• Needs clarification -- ask the user
• Low frequency / unclear -- note for later

7. Write Context Files

From everything gathered, create:

CLAUDE.md (hot context, ~60-120 lines):

# Context

## Project
- [What this repo is for]

## Current Focus
- [What matters right now]

## Always / Never
- [Few rules that matter in most sessions]

## Key Paths
- `AGENTS.md`
- `TASKS.md`
- `context/`
- `docs/`

## Context Placement
- context-advisor owner path
- repo profile

context/ directory:

• context/glossary.md -- full decoder ring, acronyms, terms, nicknames, codenames
• context/people/{name}.md -- individual profiles
• context/projects/{name}.md -- project details
• context/reference/company.md -- teams, tools, processes

Do not create both CLAUDE.md and .claude/CLAUDE.md at the same repo root.

Keep this context local to the owning repo by default. Use context-advisor before promoting durable cross-context knowledge.

8. Report Results

Productivity system ready:
- Connectors: .productivity.yml (calendar: X, email: X, tracker: X, kb: X, chat: X)
- Tasks: TASKS.md (X items)
- Context: X people, X terms, X projects

Use /productivity-sync to keep things current (add --deep for a comprehensive scan).

Notes

• If context is already initialized, this just shows current config
• Nicknames are critical; always capture how people are actually referred to
• If a source isn't available, skip it and note the gap
• Context grows organically through natural conversation after bootstrap