mumro-academy

MUMRO Academy · academy.mumro.io

EN: Open engineering and marketing handbook for living brand systems, multi-channel content operations, technical SEO, and Model Context Protocol (MCP) AI agents — built by MUMRO (RolePlayingTech).

PL: Otwarty podręcznik inżynieryjno-marketingowy o żywych systemach marki, wielokanałowych operacjach treści, technicznym SEO oraz agentach AI i Model Context Protocol (MCP) — tworzony przez MUMRO (RolePlayingTech).

---

English Overview

Most marketing content fails not because teams lack writing tools, but because context resets at every step:

1. Brand strategy is locked inside a forgotten PDF deck.
2. Writers and AI chat tabs start from isolated, ungrounded prompts.
3. The same caption is copy-pasted across Instagram, LinkedIn, X, and Facebook.
4. Website CMS publishing and social distribution operate on separate islands.
5. Analytics dashboards report vanity likes without ever informing the next strategy revision.

MUMRO Academy (academy.mumro.io) teaches portable engineering and editorial methods you can apply in your own stack, pairs them with four independent open-source CLI kits for coding-agent and MCP workflows, and shows how MUMRO implements the connected brand, publishing, and governance loop in practice.

Curriculum: 4 Learning Tracks & 16 Deep-Dive Lessons

Track	Lesson	Title (EN)	Focus
Track 01Living Brand Systems	01	Versioned Brand Strategy: Why Static PDFs Fail Modern Content Teams	Strategy v1 → v2, pillars, channel roles & 30-day hypotheses
Track 01Living Brand Systems	02	Brand Voice Architecture: Encoding Tone, Boundaries & Vocabulary	Cadence rules, domain lexicon & negative claim boundaries
Track 01Living Brand Systems	03	Evidence-Led Research Pipelines: RSS, Official Domains & Provenance	Separating raw facts from editorial interpretation (Source → Decision → Draft)
Track 02Multi-Format Ops	04	One Story, Four Channels: Adapting Core Narratives Without Copy-Paste	Canonical website article + 5-slide carousel + IG / FB / X / LinkedIn
Track 02Multi-Format Ops	05	High-Retention Carousels: Structural Pacing from Hook Slide to Action	The 5-slide arc (Hook → Breakdown → Mechanism → Proof → Takeaway)
Track 02Multi-Format Ops	06	Publishing Operations: Queues, Time Slots, Idempotency & Receipts	Why queued ≠ published, per-target receipts & idempotent retries
Track 03Website, SEO & Loops	07	Website + Social as One Loop: Connecting CMS with Social Distribution	WordPress / API CMS delivery, canonicals, reciprocal hreflang & JSON-LD
Track 03Website, SEO & Loops	08	Grounded Website Assistants: Turning Visitor Questions into Content & Leads	Brand-grounded chatbot, Unified Inbox, lead capture & FAQ Mining
Track 03Website, SEO & Loops	09	Cohort Analytics Over Vanity Metrics: Closing the Loop to Strategy v2	Evaluating 7/30/90-day cohorts, content reuse & Strategy v2 drafts
Track 04Agents, MCP & Control	10	Model Context Protocol (MCP) in Practice: Connecting AI Agents to Brand Context	Four-tier, workspace-bound MCP tool surface for Claude Code, Codex & Antigravity
Track 04Agents, MCP & Control	11	Prepare → Confirm: Designing Safe Mutation Gates for Autonomous Agents	Two-phase confirmation tokens preventing accidental live publications
Track 04Agents, MCP & Control	12	Bounded Autonomy: Credit Budgets, Quiet Hours, Channels & Kill Switches	draft_only, schedule_with_confirmation, autonomous_schedule & RBAC
Track 04Agents, MCP & Control	13	What Is an AI Harness? From First Steps with Coding Agents to Harness Engineering	AGENTS.md, progressive-disclosure skills, verification loops & agent-harness-kit
Track 04Agents, MCP & Control	14	MCP, Tools and Skills: Why Agent-Operable Products Will Win	Four-tier MCP surfaces, per-operation scopes, untrusted data & mcp-surface-kit
Track 04Agents, MCP & Control	15	Playwright & Multimodal Browser Verification: Giving AI Coding Agents Eyes, Hands, and Deterministic UI Contracts	ARIA snapshots vs. VLM screenshots, ui-contract isolation, token-rotation session safety & playwright-harness-kit
Track 04Agents, MCP & Control	16	Git Isn’t a Save Button: Version Control, Worktrees, and Safe Recovery in the Age of AI Coding Agents	Three Trees, branches, git worktree isolation, catching weakened tests, safe recovery & git-agent-workflow-kit

Operational Playbooks, Glossary & Open-Source CLI Kits

• Playbooks (/playbooks/, /pl/playbooks/): 8 copy-ready YAML/JSON specifications for Brand Voice & Claim Boundaries, One-Story 4-Channel Decomposition, MCP Agent Campaign Loop Runbooks, Workspace Autonomy Governance, Agent Harness Blueprints (agent-harness-kit), MCP Surface Specs (mcp-surface-kit), Playwright Browser Verification Contracts (playwright-harness-kit), and Multi-Agent Git Workflow Policies (git-agent-workflow-kit).
• Glossary (/glossary/, /pl/glossary/): Structured A–Z technical & marketing definitions with DefinedTermSet schema.org markup.
• Open-Source CLI Kits (4 tools):
  • RolePlayingTech/agent-harness-kit — Scan, design, explain, build, and check coding-agent harnesses (.harness/spec.json → AGENTS.md, CLAUDE.md, .cursor/rules, and skills).
  • RolePlayingTech/mcp-surface-kit — Explain, validate (--strict), and generate SKILL.md workflows from a declarative MCP product surface contract (.mcp-surface/spec.json).
  • RolePlayingTech/playwright-harness-kit — Explain, validate (--strict --verify-files), and generate SKILL.md browser verification workflows from a declarative Playwright contract (.playwright-harness/spec.json).
  • RolePlayingTech/git-agent-workflow-kit — Read-only repository diagnostics (doctor), pre-PR scope and test-integrity audits (pr-ready), non-destructive recovery plans (recovery-plan), 5 portable Agent Skills (SKILL.md), and local Git lab sandboxes (sandbox).

---

Przegląd po polsku (Polish Edition)

Większość działań content marketingowych przegrywa nie z braku narzędzi do pisania, lecz dlatego, że kontekst marki zeruje się na każdym etapie pracy:

1. Strategia marki leży w nieotwieranym od miesięcy pliku PDF.
2. Copywriterzy i czaty AI startują od pustego okna bez znajomości źródeł i granic marki.
3. Ten sam tekst jest mechanicznie przeklejany na Instagrama, LinkedIna, X i Facebooka.
4. Strona WWW (CMS) oraz kanały społecznościowe działają w całkowitej izolacji.
5. Wykresy analityczne nie przekładają się na aktualizację strategii w kolejnym miesiącu.

Akademia MUMRO (academy.mumro.io/pl/) to otwarty, dwujęzyczny podręcznik pokazujący uniwersalne metody inżynieryjne i redakcyjne, które wdrożysz we własnym stosie narzędzi, cztery niezależne pakiety CLI open source dla warsztatu agentów kodujących i MCP oraz praktyczne przykłady tego, które elementy obiegu treści realizuje MUMRO.

Program: 4 Ścieżki, 16 Lekcji, 8 Playbooków i 4 Otwarte Narzędzia CLI

1. Ścieżka 01 — Żywe systemy marki i strategia treści: Wersjonowana strategia (v1 → v2), inżynieria Brand Voice i twardych granic marki oraz research oparty na dowodach (Źródło → Decyzja → Szkic).
2. Ścieżka 02 — Wielokanałowa inżynieria treści i publikacja: Rozwijanie jednej historii na artykuł WWW, karuzelę 5 slajdów i 4 kanały społecznościowe oraz odporna kolejka publikacji z potwierdzeniami per kanał.
3. Ścieżka 03 — Strona WWW, techniczne SEO i pętla rozmów: Integracja CMS (WordPress / API), dwujęzyczne SEO, osadzone w wiedzy czatboty WWW, wspólny inbox, FAQ mining i analityka kohortowa 7/30/90 dni.
4. Ścieżka 04 — Agenci AI, Model Context Protocol (MCP) i kontrola człowieka: Łączenie agentów AI z kontekstem marki przez MCP, dwufazowe bramki decyzyjne Prepare → Confirm, 3 poziomy autonomii, inżynieria harnessu agenta kodującego (agent-harness-kit), projektowanie powierzchni produktów MCP (mcp-surface-kit), weryfikacja przeglądarkowa Playwright (playwright-harness-kit) oraz wieloagentowy workflow Git (git-agent-workflow-kit).

---

Design System, Technical SEO & Architecture

This repository is engineered with the exact visual and technical standards of mumro.io:

• Editorial-Technical Aesthetic: Warm paper (#f5f3ed), deep ink (#20231f), signal accent (#e7471b), Instrument Serif display typography, Inter body copy, and Cascadia Mono technical metadata—including full Polish diacritic subsets (woff2).
• Deterministic Bilingual Static HTML: Built with Astro 7 (output: 'static'). Every route exists in both English (/, /tracks/.../, /lessons/.../, /playbooks/, /glossary/) and Polish (/pl/, /pl/tracks/.../, /pl/lessons/.../, /pl/playbooks/, /pl/glossary/).
• Compile-Time Locale Parity: assertLocaleParity() validates that English and Polish content trees share identical keys, array lengths, and non-empty strings at build time.
• Technical SEO & LLM Readiness:
  • Self-referencing <link rel="canonical"> and reciprocal <link rel="alternate" hreflang="en|pl|x-default"> on every route.
  • Rich JSON-LD @graph (WebSite, EducationalOrganization, Course, TechArticle, LearningResource, BreadcrumbList, DefinedTermSet, FAQPage).
  • Generated 1200×630 Open Graph / Twitter share cards (/media/share-en.png, /media/share-pl.png).
  • XML Sitemap (/sitemap.xml) with xhtml:link alternates, /robots.txt, bilingual RSS 2.0 feeds (/rss.xml, /pl/rss.xml), /llms.txt, and /llms-full.txt (complete plain-text curriculum for AI indexers & coding agents).
  • Interactive System Labs (LessonLab.astro): Embedded interactive simulators on every track and lesson page for inspecting Brand Voice & Source Provenance layers, 5-slide carousels, 7/30/90-day cohort windows, and two-phase MCP Prepare → Confirm gates.

Repository Structure

mumro-academy/
├── deploy/
│   └── academy.mumro.io.nginx.conf  # Nginx config with EN/PL negotiation & caching
├── public/
│   ├── favicon.svg
│   └── media/                       # Generated 1200x630 EN/PL Open Graph share cards & SVGs
├── scripts/
│   ├── generate-share-images.mjs
│   ├── package-release.mjs          # Deterministic release archive & SHA256 manifest builder
│   ├── share-fonts.css
│   └── verify-build.mjs             # Automated check of all 46 routes, SEO tags, JSON-LD & feeds
├── src/
│   ├── assets/fonts/                # OFL Latin + Polish woff2 font subsets
│   ├── components/                  # AcademyHome, TrackPage, LessonPage, LessonLab, PlaybooksPage, GlossaryPage
│   ├── content/                     # Bilingual EN/PL tracks, 16 lessons, 8 playbooks, glossary, tools & shell copy
│   ├── layouts/BaseLayout.astro
│   ├── lib/contracts.ts             # Locale types, URL helpers & assertLocaleParity
│   ├── pages/                       # Static EN & PL routes + sitemap.xml, robots.txt, llms.txt, llms-full.txt, rss.xml
│   ├── scripts/shell.ts             # Progressive enhancement: filters, labs, loop explorer, copy, checklist, dialog
│   └── styles/                      # fonts.css & global.css (MUMRO design tokens)
└── tests/
    └── academy.spec.ts              # Playwright E2E, bilingual parity, interactive labs & responsive tests

Local Development & Verification

# Install dependencies
npm install

# Start local dev server (http://127.0.0.1:4325)
npm run dev

# Run Astro & TypeScript strict type-checking
npm run check

# Build static production output + run automated SEO & route verification
npm run build

# Run Playwright browser tests
npm test

Links

• MUMRO Academy: https://academy.mumro.io · https://academy.mumro.io/pl/
• MUMRO Platform: https://mumro.io · https://mumro.io/pl/
• RolePlayingTech: https://roleplayingtech.com