Why Claude

Why Claude is the one I reach for

Reads the whole room

Long context means whole transcripts, PRDs and research dumps go in at once, and answers can point back to the line they came from.

How a request moves through the kit

Your inputClaudeSkillSub-agents, in parallelOutcomes
A one-line pitch from a Slack thread
Claude
Synthesis + self-audit
Five-pillar scorecardHTML
Six-month pre-mortemDoc
Go / no-go test planDoc
Skills

The skills, a playbook for every part of the job

All 54 skills in the kit, grouped by the kind of work they support. Pick a theme to see its skills.

Discovery, Intelligence & Synthesis

Raw inputs like call notes, Slack threads, rival pages and half-formed ideas, turned into decisions and evidence before anything gets written.

  1. 01
  2. 02
  3. 03
  4. 04
  5. 05

Discovery, Intelligence & Synthesis

Raw inputs like call notes, Slack threads, rival pages and half-formed ideas, turned into decisions and evidence before anything gets written.

  1. 01
  2. 02
  3. 03
  4. 04
  5. 05

Prioritization, Roadmapping & Commercial Modeling

When everything is P0. Scoring, cost of delay, unit economics and roadmaps that survive a leadership review.

  1. 06
  2. 07
  3. 08
  4. 09
  5. 10
  6. 11
  7. 12
  8. 13

Documentation, Requirements & Execution

From a validated idea to a buildable spec: PRD, FRD, DRD and stories, with IDs that trace end to end.

  1. 14
  2. 15
  3. 16
  4. 17
  5. 18
  6. 19
  7. 20
  8. 21
  9. 22

UX, Service Design & Design Governance

Jobs, flows, blueprints and design audits, so reviews start from the user’s mental model, not the schema.

  1. 23
  2. 24
  3. 25
  4. 26
  5. 27
  6. 28
  7. 29
  8. 30

Growth, PLG & Product Analytics

North Star trees, guardrails, experiments and activation, so every launch ships instrumented.

  1. 31
  2. 32
  3. 33
  4. 34
  5. 35
  6. 36
  7. 37
  8. 38

AI Product Management

For AI features specifically: whether it is a fit, how retrieval works, what it costs and how it fails safely.

  1. 39
  2. 40
  3. 41
  4. 42
  5. 43

Storytelling, GTM & Lifecycle

Narratives, buy-in, launches and sunsets, the work after the spec that decides whether it lands.

  1. 44
  2. 45
  3. 46
  4. 47
  5. 48
  6. 49
  7. 50
  8. 51
  9. 52
  10. 53
  11. 54
Sub-agents

The sub-agents, a whole product team on call

All 24 sub-agents in the kit: builders, stakeholders and a panel of synthetic users and buyers. Each skill calls the few it needs.

~/.claude/agents24 items

Squad AProduct, Design & Engineering

  • Used by 32 skills
  • Used by 25 skills
  • Used by 24 skills
  • Used by 12 skills
  • Used by 17 skills
  • Used by 17 skills
  • Used by 5 skills

Squad BGTM, Operations & Leadership

  • Used by 15 skills
  • Used by 15 skills
  • Used by 8 skills
  • Used by 15 skills
  • Used by 5 skills
  • Used by 14 skills
  • Used by 5 skills
  • Used by 1 skills
  • Used by 8 skills

Squad CSynthetic User & Buyer Panel

  • Used by 12 skills
  • Used by 13 skills
  • Used by 7 skills
  • Used by 7 skills
  • Used by 6 skills
  • Used by 3 skills
  • Used by 4 skills
  • Used by 10 skills
Setup

How a skill runs

  1. 1

    Read the source

    Works out what it was handed (a transcript, PRD, screenshot or pitch) and extracts signals with rules written for that input.

    source-types.md
  2. 2

    Ask only the gap

    Three to six questions it can’t infer, with an explicit list of what never to ask because the input already answers it.

    follow-up-protocols.md
  3. 3

    Pick the format

    Chat, an interactive HTML artifact, .docx, .xlsx, or a straight push into Jira or Notion.

    output-formats.md
  4. 4

    Dispatch and audit

    The squad reviews in parallel, the skill synthesises their verdicts, then grades its own draft against a named rubric.

    sub-agent-dispatch.md

Take what’s useful

In Claude

  1. Turn on code execution and file creation in Settings → Capabilities, if it isn’t already.
  2. Go to Settings → Customize → Skills, hit +, then Create skill → Upload a skill.
  3. Upload the skill’s .skill file.
  4. Hand Claude the input it expects. The skill triggers from its description.

In Claude Code

# skills — ~/.claude for everywhere, .claude for this repo
unzip pm-prd-builder.skill -d ~/.claude/skills/

# sub-agents — same choice, project wins on a name clash
cp ux-designer.md qa-lead.md ~/.claude/agents/

Both are picked up on the next session, no restart needed. Call a sub-agent by name with @ux-designer.

The whole kit, one zip.

  • 54 skills
  • 24 sub-agents
  • 3 protocols
  • 1 README

AI-native PM

A PM doesn’t just use AI.
He works as a team with it.

Tools he works with

  • ClaudeReasoning and agentic building
  • Google AI StudioMultimodal prototyping
  • MistralOpen-weight models
  • LovableRapid prototyping
  • v0 by VercelUI generation
  • Figma AIDesign and prototyping
  • SupabaseAI-ready backend
  • n8nWorkflow automation

A PM in every seat

Product
Manager+ AI
Staff PM
Researcher
Designer
Developer
Tester
Marketer
Project Manager
Consultant

What comes out

  • Ship a full-stack product soloSpec, build and deploy without waiting on a team
  • Run discovery in a day, not a sprintRivals, pricing and user evidence, synthesised
  • Design a modern UI without a designerFlows, states and screens, production-ready
  • Turn a journey into a groomed backlogINVEST epics and stories, pushed to Jira
  • Catch the edge cases before QA doesThree-lens risk sweep with a running register
  • Automate the whole GTM launchMessaging, runbook, battlecards and rollback
  • Win the room with a six-page narrativeExec decks and buy-in cases that survive review
  • Put the repetitive half on autopilotStatus, grooming prep and reporting, handled