By sethshoultes
Multi-agent AI agency — deploy a team of AI personas (Steve Jobs, Elon Musk, Marcus Aurelius, Jensen Huang + design crew) that debate strategy, hire sub-agents, build deliverables, write code, and ship products from a single PRD.
Use this agent for video scripts, demo walkthroughs, tutorial content, launch videos, and workshop materials. Aaron Sorkin brings razor-sharp dialogue, precise pacing, and the rare ability to make technical work feel urgent and alive. Use when you need to explain something visually, create compelling video content, or turn a dry product walkthrough into something people actually watch to the end. Examples: - User: 'Write a product demo script for our dashboard feature' Assistant: 'Sorkin will make the demo feel like a scene — the user has a problem, the product solves it, and every click has a reason.' - User: 'Script a 60-second launch video' Assistant: 'Sorkin hooks in the first three seconds, builds to a moment of clarity, and lands with something that sticks.' - User: 'Write the tutorial for our onboarding flow' Assistant: 'Sorkin writes tutorials as conversations — the viewer is always one step ahead of where they thought they were going.'
Implement backend logic — APIs, services, business logic, data access, integrations. Use when a director (Elon, Steve, Margaret) dispatches implementation work that involves server-side code, REST/GraphQL endpoints, business logic, queue handlers, scheduled jobs, or third-party API integrations. Returns production-ready code with appropriate error handling and minimal scope. Functional-role implementer (Sonnet — code writers stay on Sonnet for craft accuracy) — no biographical persona, just craft and conventions.
Review code for craft, convention, and obvious correctness. Use when Margaret, Steve, or Elon dispatches a pre-merge review. Returns a structured findings list — what's good, what's wrong, what's worth flagging. Functional-role implementer (Haiku tier).
Design and modify database schemas, write migrations, optimize queries, manage indexes. Use when Elon dispatches data-layer work — new tables, schema changes, query performance, indexing strategy, or migration planning. Returns production-grade schema with migration scripts and rollback plan. Functional-role implementer (Sonnet — code writers stay on Sonnet for craft accuracy).
Implement CI/CD, infrastructure, observability, deploys. Use when Elon or Margaret dispatches infra work — GitHub Actions workflows, deploy scripts, environment setup, monitoring/logging wiring, cron jobs, container configs. Returns production-grade configs with clear rollback paths. Functional-role implementer (Sonnet — code writers stay on Sonnet for craft accuracy).
Query and manage the Great Minds agency vector store — semantic search over memories stored in SQLite with embeddings. Subcommands - search, add, list, import, export, stats. Filters by type (learning, decision, board-review), agent, project. Use to recall prior decisions, learnings, or critiques across multiple agency projects.
Structured phase planning with XML task plans. Researches the codebase, generates atomic task plans in XML format, verifies them against requirements, and organizes them into dependency-aware execution waves. Inspired by GSD (Get Shit Done) methodology.
Compare current work against the original plan to detect scope creep. Reads .planning/*-PLAN.md or prds/*.md, checks git diff --stat and recent commits, maps each changed file to a planned deliverable, flags unplanned work, and reports drift percentage. Use during execution when context-guard signals drift, or run manually to keep a project on-spec.
Launch the Great Minds agent swarm using Agent tool with worktree isolation. No tmux dispatch — proven unreliable. Phil orchestrates from admin, spawns sub-agents via Agent tool.
Generate or regenerate the .wolf/anatomy.md file index with token estimates for the current project.
Uses power tools
Uses Bash, Write, or Edit tools
Own this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimOwn this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimBased on adoption, maintenance, documentation, and repository signals. Not a security audit or endorsement.
14 legendary personas (Jobs, Musk, Buffett, Ive, Rubin, Huang, Winfrey, Rhimes, Blakely, Hamilton, Angelou, Sorkin, Aurelius, Jackson) plus 8 functional-role implementers — that debate, review, plan, and build. Available in three install formats to match how you work.
📖 Read the User Manual for the complete reference — install paths, all 22 agents (14 named + 8 functional), all 17 skills, lite vs full agency workflows, project structure, memory store, three end-to-end walkthroughs, patterns, and troubleshooting.
/constellation-start
The discoverable entry point for the Caseproof persona constellation. Asks 2-3 questions about your project shape and routes to the right plugin's project-init skill — or channels Phil Jackson (constellation-aware as of v1.3) for ambiguous or mixed projects. Always ensures the bible at .great-authors/project.md exists, since that's the shared spine across all five plugins.
If you already know which plugin you want, you can skip the entry point and run that plugin's project-init directly. /constellation-start exists for the case where you don't.
Polish from the cross-plugin quality review. No new user-facing features; everything is hygiene and parity with the rest of the constellation.
tests/smoke.sh — first smoke test for great-minds. Validates SKILL.md frontmatter, persona frontmatter (name / description / model / color), DXT persona-subset membership against agents/, version coherence across package.json / plugin.json / DXT manifests / DXT server self-report, DXT tool-handler alignment, and v1.3+ /constellation-start skill presence.LICENSE — MIT, matching the rest of the constellation. The plugin had been shipping without one.agency-daemon, scope-check, agency-memory.color: field added to aaron-sorkin-screenwriter.md (the only persona missing it).distribution/plugin/skills/constellation-start/ — the lite distribution now includes the constellation entry point. Newcomers using the lite install can run /constellation-start without falling back to the full bundle. Phil Jackson's persona file in lite is also resynced with the constellation-aware expansion from v1.3.A cross-plugin review pass surfaced these as accumulated drift. None were user-blocking, but together they were a real consistency gap — every other plugin in the constellation had smoke tests, and great-minds was the most likely to drift without them given its breadth (22 personas, 20 skills, infrastructure for a swarm runtime).
Constellation-level entry point. The plugin now has a discoverable starting point for newcomers to the entire Caseproof persona constellation (great-minds + great-authors + great-filmmakers + great-publishers + great-marketers — five plugins, each owning one craft).
/constellation-start (new skill) — asks project-shape questions, routes to the right plugin's project-init or channels Phil Jackson for mixed projects. Discoverable, lightweight, ~150 lines.phil-jackson-orchestrator.md (expanded) — Phil's existing role (Zen-master swarm orchestrator within great-minds) is preserved unchanged; a new "Your Role at the Constellation Level" section adds the orchestration knowledge for cross-plugin work. Same triangle-offense logic; one level up. Eleven rings, not one.constellation_start tool. Available in Claude Desktop alongside the existing tools.Doesn't break anything. /agency-start and the entire agency-* swarm-pattern family stay intact — they're the right answer for software-product swarm projects, and /constellation-start dispatches to them when the user picks that branch.
The three-layer architecture is now explicit and shipping. Grounded in 2026 research (Wharton, USC) on persona prompting — named expert personas excel at judgment but reduce factual accuracy on knowledge-heavy tasks; functional roles avoid that trap.
npx claudepluginhub sethshoultes/great-minds-plugin --plugin great-minds14 legendary personas (Jobs, Musk, Buffett, Ive, Hamilton, Angelou, Rubin, Huang, Winfrey, Rhimes, Blakely, Aurelius, Jackson, Sorkin) plus co-work skills: structured debate, full board review, content publishing pipeline, video pipeline, codebase anatomy, and scope drift check.
Skills + MCP connector for calling garagedoorscience.com. Gives Claude instant diagnostic and routing tools for residential garage doors.
Nine researcher personas (Sagan, Gould, Roach, Sacks, Gawande, Diamond, Wilson, Skloot, Caro) plus four operational skills for studies, peer reviews, and project initialization. Drafted by great-authors writers via cross-plugin orchestration. Tenth and final v0.1 plugin in the Caseproof persona constellation. NOT ACADEMIC ADVICE — a craft register. For technical-mathematical writing rigor, cross-dispatch great-engineers:don-knuth-engineer; for political-philosophy register, cross-dispatch great-counsels:hannah-arendt-counsel.
Nine operations personas (Cook, Grove, Munger, McCord, Deming, Ohno, Horowitz, Walton, Kelleher) plus four operational skills for operating plans, process reviews, and project initialization. Drafted by great-authors writers via cross-plugin orchestration. Eighth in the Caseproof persona constellation. Warren Buffett stays in great-minds for strategic capital allocation (cross-dispatchable); operators handles execution craft.
Multi-agent AI agency — deploy a team of AI personas (Steve Jobs, Elon Musk, Marcus Aurelius, Jensen Huang + design crew) that debate strategy, hire sub-agents, build deliverables, write code, and ship products from a single PRD.
Multi-agent team presets with 6 agents, 5 skills, and 3 team workflows for idea validation, PRD review, and competitive analysis
Chaotic multi-persona AI discovery system. 16 unique Vern personas, VernHole council tiers, Oracle vision, and a full multi-LLM discovery pipeline.
Compile humans into AI agents through deep interviews. Conducts comprehensive behavioral interviews, analyzes work artifacts via MCP tools, and generates installable Claude Code plugins that embody a person's decision-making, communication style, and expertise.
Adversarial thinking partner for founders and executives. Stress-tests plans, prepares for board meetings, navigates hard decisions, and forces honest post-mortems.
Dynamically assemble expert agent teams for complex tasks using Claude Code's agent teams feature
Build and run multiple personas across any AI agent.