SKILL.md files teach the AI the structure a senior professional actually uses — quality checks and anti-patterns included — so the first draft is one you can ship.npx pm-claude-skills add --agent claude| 🏛 The Boardroom | Five AI executives debate your PRD in rounds — severity-tagged openings, cross-examination, a Chair's verdict memo with the changes that would survive the room. 🔊 Theatre mode: they argue out loud. Verdicts export as hash-verifiable attestations. |
| 🏢 The Firm | Charter a standing AI staff (CFO, CCO, CTO, Strategy…) who file delta-aware memos, hold board meetings without you, and log falsifiable predictions you score — each staff member earns a visible calibration record. |
| 🏆 The Gauntlet | Job posting → decoded role → your stories forged → live cross-examination (answer by voice) → salary negotiation vs a hidden band → readiness /100 + a frameable certificate. |
| 🩻 Document X-ray | Every sentence heat-mapped by evidence strength — data · sourced · assertion · superlative · hidden assumption — load-bearing lines outlined, weakest claims listed to fix first. |
| 🎛 Living artifacts | RICE renders as a live board — drag the confidence slider, the ranking re-sorts. Journeys draw their emotion curve; health checks animate gauges. |
| 🥊 The Gym | Negotiate a salary / renewal / vendor deal against an AI with a secret bottom line; the debrief reveals what was achievable and scores you against it. |
| You need… | Skill |
|---|---|
| Product docs that ship | prd-template rice-prioritisation okr-builder |
| Engineering rituals | incident-postmortem architecture-decision-record ai-code-review |
| Exec & board comms | executive-update stakeholder-update |
| Customer success | cs-health-scorecard churn-analysis renewal-playbook |
| The agent era 🆕 | mcp-server-spec agent-readiness-audit agent-era-pricing |
| AI at work 🆕 | ai-usage-policy ai-roi-audit role-redesign-for-ai |
| Your career | pm-personal bundle + the Gauntlet |
references/ (the judgment calls) + templates/ (fill-in worksheets). 🇪🇸 The whole tier is in Spanish too.| Claude Code (+4 agents, 21 commands, hooks) | npx pm-claude-skills add --agent claude |
| Cursor / Codex / Windsurf / 60+ | … add --agent cursor |
| Any MCP client | claude mcp add pm-skills -- npx -y pm-claude-skills-mcp |
| ChatGPT / Claude.ai connector | pm-skills-mcp.pm-claude-skills.workers.dev |
| Python (LangChain/CrewAI) | pip install pm-skills |
| Plugin marketplace | /plugin marketplace add mohitagw15856/pm-claude-skills |
| GitHub CI (PR descs, changelogs, doc lint) | copy a file from action/examples/ |
| Browser extension (insert skills + lint your writing) | extension/ |
/ship-a-feature — fuzzy idea → problem → PRD → RICE → roadmap → GTM/land-a-job — decode → research → apply → interview/ship-an-mcp-server 🆕 — spec → readiness audit → pricing → oversight/adopt-ai-properly 🆕 — policy → roles → reviews → ROI audit/launch-a-product, /close-the-quarter, /rescue-an-account, /run-discovery[data]→[hunch]; skills read it before answering, write back approval-gated. No vector DB.outcome-tracker — every decision's predictions recorded & scored against reality; your frameworks earn calibration records.action-runner — recommendations become real tickets/messages, risk-rated, approved per action. Nothing acts silently.SKILL.md standard (conformance L1–L4); SkillCheck + security audit gate every PRyou/skill-name via one PR; served live over REST + MCPmessage/send; MCP pairings make skills act (Notion, Slack — approval-gated)/v1/skills · /v1/search · /v1/workflows · /v1/community (no auth) → n8n, Lovable, Obsidian