Arcanium

What's Inside

• /mine — extract & route decisions, actions, issues, risks from any transcript
• /review — weekly vault health check
• /do — action item dashboard
• /issues — issue tracker with ISS-NNN numbering
• /consult — structured decision brief
• /learn — web research + vault routing
• 3 hooks — session-orient, auto-commit, write-validate
• 9 schema templates — decisions, actions, issues, risks, strategy, competitors, pipeline, OKRs, archive
• Meridian Analytics worked examples included

Built For

• Founders who need to trace why a decision was made
• Consultants who run meetings and document them
• Teams where "who owns this?" is a daily question
• Claude Code users who want AI with org context

Requirements

• Obsidian (free)
• Claude Code (Anthropic subscription)
• Git (free) or Obsidian Sync

Setup time: 10 minutes flat.

What's Inside

• 7-part guide (~5,000 words) — from Agile's breaking assumptions to the four-stage transition roadmap
• New Team Topology — Product Owner · Business Analyst · Forward-Deployed Engineer
• Agent-Ready Work Spec — the 5-component format that replaces Jira tickets for agent execution
• FDE Context Package Template — the CONTEXT.md your agents receive before every task
• SPEC-NNN Template — numbered work specs with FDE review notes section
• Team Topology Cheat Sheet — roles, review gate, transition stages, tool stack

Who It's For

• Engineering managers navigating the AI-agent shift
• Agile coaches ready to tell clients where teams are heading
• Technical leads doing the FDE job without a framework to make it legible
• CTOs and founders building small, fast teams from scratch
• Anyone whose sprint ceremonies are producing less value

Key Concepts

• Jidoka applied to agent review — Toyota's "stop and fix" principle for the FDE review gate
• Stage 1→4 transition roadmap — where most teams are and what Stage 4 looks like
• Rickover's accountability doctrine — you may delegate authority; you may not delegate responsibility

What's Inside

• Why local AI — and why Apple Silicon
• Hardware tiers: every current Mac + iPad Pro model mapped
• Ollama: install, CLI, API, essential commands
• LM Studio: GUI model management
• iPad Pro: Private LLM, LLM Farm, Enchanted apps
• Image generation: Draw Things + FLUX.2 on Mac and iPad
• Model selection by task and by hardware tier
• Model routing cheat sheet — task-to-model quick reference
• Troubleshooting: memory, speed, and API issues

Built For

• Mac + iPad Pro owners who want offline AI
• Developers avoiding external API costs and data exposure
• Privacy-conscious users keeping prompts on-device
• Anyone evaluating Apple Silicon AI capabilities

Requirements

• Apple Silicon Mac (M1 or newer) or iPad Pro M4/M5
• 10–100 GB free disk space (model dependent)
• All tools covered are free to download

Setup time: under 20 minutes.

What's Inside

• 14-page PDF guide — six chapters: how Claude Code works, CLAUDE.md, hooks, skills, context management, MCP
• Quick-reference card — single-page landscape cheat sheet, print it and pin it
• session-orient.sh — git status, open tasks, recent commits at session start
• auto-commit.sh — every write goes to git automatically with a descriptive message
• write-validate.sh — markdown frontmatter validation before Claude writes
• session-log.sh — timestamped audit trail of every tool call

Who It's For

• Claude Code users who've set up hooks but aren't sure they work
• Developers standardizing Claude Code across multiple projects
• Anyone who's lost context mid-session and wants a systematic fix
• Teams building shared CLAUDE.md conventions

What You'll Learn

• Exit 0 vs. exit 1 in hooks — when to block Claude and when to fail silently
• Skill invocation patterns — natural language only; why /skillname never works
• The orient-first pattern — keeping context lean across long sessions
• Task tool for subagents — spawning parallel agents without cluttering main context

What's Inside

• 5-part strategic guide (~5,000 words) — three layers, six decisions, vendor evaluation, CIO conversation, 90-day checklist
• The Six Decisions — build/buy/wait · infrastructure vs. application layer · vendor lock-in · governance · pilot selection · workforce transition
• Vendor Filter — four questions that separate real enterprise deployments from vaporware pitches
• CIO Brief Template — one-paragraph executive conversation starter, ready to adapt and send
• 90-Day Decision Checklist — four phases, concrete deliverables, pilot framed and authorized
• Quick-reference card — all six decisions, vendor filter, CIO brief, and sprint checklist on four pages

Who It's For

• Technical leaders responsible for enterprise IT strategy
• Engineering leads navigating agent platform vendor pitches
• CTOs and VPs evaluating whether OpenClaw belongs in the roadmap
• Anyone whose CIO just asked "what's our agent orchestration strategy?"
• Organizations piloting AI agents without a governance framework yet

Key Concepts

• The Pilot-Forever Trap — why experiments without graduation criteria aren't strategy
• Infrastructure vs. Application Layer — most orgs don't need an OpenClaw decision yet
• The Vendor Gap — why established relationships are selling last cycle's technology
• Jidoka applied to agent governance — quality built in at the review gate, not inspected out after