Skip to content
shell
$ npx -y skills add aleksander-dytko/ai-pm-workspace --skill decision --agent claude-code

How it fires

How this skill gets triggered: by you, by Claude, or both.

  • Fires itselfAuto-invocation. Claude auto-loads it when your prompt matches the work.
  • You can call itInvoke it directly when you want it.
  • Slash command/decision
How auto-invocation works

Context preview

The summary Claude sees to decide when to auto-load this skill.

Make a product decision with full context from the vault and available sources

SKILL.md

decision.SKILL.md
name: decision
description: Make a product decision with full context from the vault and available sources
argument-hint: "[decision topic]"

Make a Product Decision

You help make a well-informed product decision by gathering context from all available sources and producing structured documentation, follow-up tasks, and a communication draft.

Input

The user provides a decision topic via `$ARGUMENTS`. Examples:

  • "should we support feature X in the next release"
  • "prioritize feature A vs feature B"
  • "should we adopt GraphQL for the new API"

Workflow

1. **Search the vault**:

  • Recent meetings in `Meetings/` that touched this topic
  • Daily notes in `journals/` (last 2 weeks) for related context
  • `Loose Notes/Work/` for related decisions or drafts
  • `Dashboard/Weekly P-Tasks.md` for related priorities

2. **Use GitHub MCP** (if configured):

  • Search repositories for related issues/PRs
  • Read engineering comments and discussions
  • Identify any blockers or concerns raised

3. **Ask for additional context** (before options):

  • "Do you have a Slack thread, document, or discussion about this decision? If yes, paste it now."
  • If provided: extract opinions, concerns, constraints.
  • If skipped: proceed without.

4. **Present context summary**:

  • Relevant notes and prior discussions
  • Strategy alignment (if strategy docs exist)
  • Customer data or validation (if any)
  • Engineering discussions and feasibility (from GitHub)

5. **Draft options with trade-offs**:

  • 2-3 viable options
  • For each: pros (benefits, alignment, customer value), cons (costs, risks, complexity, timeline)
  • Do not recommend yet - present objectively

6. **Create decision note**:

  • Location: `Loose Notes/Work/YYYY-MM-DD - Decision - [Topic].md`
  • Structure:
---
tags: LooseNotes
date: YYYY-MM-DD
---
# Decision: [Topic]

**Date**: YYYY-MM-DD

## Context
[Why this decision is needed - background from notes, meetings, strategy]

## Options Considered
1. **Option 1**: [Description]
   - Pros: ...
   - Cons: ...
2. **Option 2**: [Description]
   - Pros: ...
   - Cons: ...

## Decision
[To be filled after review, or suggest recommendation if clear]

## Rationale
[Why this option, trade-offs accepted]

## Impact
[What changes, who is affected, timeline]

## Communication
[Who needs to be informed, how]

## Follow-up Tasks
- [ ] [Task]

7. **Link the decision note**:

  • Add a link in today's journal (`journals/YYYY/MM-Month/DD-MM-YYYY.md`) under `## Notes`.

8. **Draft a communication**:

  • A ready-to-post message announcing the decision.
  • Keep it concise (under 200 words).
  • Format: Context → Decision → Impact → Next steps.
  • Channel-neutral by default; if the user names a channel (Slack/email), adapt to that channel's style (see `/communicate`).

9. **Propose follow-up tasks** (confirm before writing):

  • Identify tasks needed to implement and communicate the decision.
  • Show the proposed list and wait for confirmation.
  • After confirmation, append each task to `Dashboard/tasks.md` under "This week" with format:
- [ ] Task - source: [[YYYY-MM-DD - Decision - Topic]] - due: YYYY-MM-DD - priority: P2

Output Format

Report results in conversational prose, not a fenced template card. Use backtick-wrapped paths for any file references in chat (wikilinks stay inside the decision note itself, not chat output).

Cover these beats in 3-4 sentences of prose:

  • Where the decision note was written (backtick path to the new file in `Loose Notes/Work/`).
  • Whether it was linked in today's journal (or "skipped - no journal for today").
  • How much context was pulled in (rough count: N related notes, N GitHub issues/PRs).
  • Your recommendation in one line - or "awaiting your input" if the decision genuinely could go either way.

Then show the communication draft (this one stays fenced - it's copy-paste content) and the proposed follow-up tasks as a numbered list so the user can approve with `y` / `n` / picking numbers.

Example tone:

> Decision note written to `Loose Notes/Work/2026-04-21 - Decision - Invoice portal scope.md`, linked in today's journal. Pulled context from 3 related meeting notes and 2 GitHub issues. My read: ship the top-5 fixes first, defer the full portal to Q3 - self-service has clearer 30-day ROI than a full portal buildout. > > Draft Slack message to #billing-team: > ``` > [message body] > ``` > > Proposed follow-up tasks for `Dashboard/tasks.md`: > 1. Write spec for top-5 complaint fixes - P2, due 2026-04-28 > 2. Book Q3 roadmap review for the portal scope - P3 > > Add these? (Y / N / pick numbers)

Notes

  • Always cite sources (meeting notes, GitHub issues).
  • If options are equal/unclear, present objectively and ask for user preference.
  • Leave the "Decision" section empty if not immediately clear.
  • If the decision is urgent, propose tasks with P2 priority.
Read more
Read it on GitHub ↗
Ships withai-pm-workspace

A structured, Claude-Code-ready workspace for product managers, designers, and other PM-adjacent roles. Your notes become the persistent memory; Claude Code becomes the executor. This is a template.

Get the whole plugin, auto-invoked
Stats
32
Stars
0
Views
5
Forks
Maintained
Maintenance
MIT
License
3mo ago
Last commit
5mo ago
Created

Repo: aleksander-dytko/ai-pm-workspace