thinktank
Think Tank
Simulate a focused, high-signal conversation between 3 domain experts discussing a topic relevant to the current project or question. The goal is amazing advice in a single shot — not a performance, but a genuine collision of perspectives that produces insights none of them would reach alone.
How It Works
- Analyze the context - Look at the current project, codebase, and user's question
- Suggest 3 experts - Propose real-world experts whose perspectives would be valuable:
- Experts should be complementary, not just diverse — they need enough shared context to genuinely engage with each other's points, but different enough frameworks that real tension emerges naturally
- Avoid picking experts from completely unrelated domains just for variety — the goal is productive friction, not parallel monologues
- Prefer practitioners over commentators — someone who built a thing in the domain is more valuable than someone who writes about the domain. Mix theorists with builders for the best tension.
- Briefly explain why each expert is relevant and what tension they bring
- Ask user to confirm or suggest alternatives
- Simulate the discussion - Create a focused back-and-forth conversation:
- Experts must genuinely disagree, challenge, and push back — not politely build on each other
- Each expert stays in character with their known philosophy
- Reference their actual work, books, or known opinions where relevant
- The conversation should have at least one moment where an expert's point forces another to genuinely reconsider or concede
- Arrive at synthesized recommendations - Lead with "The #1 Thing" — the single most important takeaway, then 2-3 supporting recommendations. All must be insights that only emerge from the collision of perspectives — not generic advice any single expert could give independently. If a recommendation could appear in a blog post by one of the experts alone, it's not synthesized enough.
More from michaelliv/dotskills
slack-cli
Interact with Slack from the command line. Send messages, read channel history, list channels and users, upload files, manage reminders, and more. Use when the user wants to communicate with Slack, check messages, or automate Slack workflows.
37jira-cli
Interact with Atlassian Jira from the command line. Create, update, and search issues, manage sprints and epics, transition statuses, and more. Use when the user wants to work with Jira tickets, check sprint progress, or automate project workflows.
8cli-design
Review and improve CLI program design using principles from clig.dev — the Command Line Interface Guidelines. Covers help text, output, errors, arguments/flags, subcommands, interactivity, configuration, naming, and distribution. Use when designing a new CLI, reviewing CLI UX, or fixing how a command-line tool communicates with users.
5handoff
Generate a prompt for a fresh conversation to continue the current implementation. Use when hitting context limits, ending a session, or needing to hand off work to another agent/conversation.
1issue-plan
Break a design goal into detailed, dependency-ordered GitHub issues. Use when the user has a feature idea, architectural change, or refactor and wants a structured plan with issues ready for sprint execution. Triggers include "plan this", "break this down", "create issues for this", or presenting a design/architecture to implement.
1issue-sprint
Rapid iteration on GitHub issues — prioritize, create worktrees, generate prompts for subagents, review PRs, merge, repeat. Use when the user wants to work through multiple issues quickly, says "let's sprint", "batch these issues", or wants to parallelize work across multiple agents.
1