Menu
Generate a large idea set with four parallel agents, then filter it down through two debate rounds. Use when the user wants to brainstorm or explore possibilities for a project, product, feature, or strategy.
Install with Codex or Claude Copy this prompt, paste it into Codex, Claude, or another assistant, and let it review the skill page and install it for you.
Based on SOC occupation classification
Analyze Claude Code session data from ~/.claude/projects/ to surface usage patterns, heavy sessions, and context-limit problems.
Draft a cold email that gets a reply from someone influential — an investor, founder, hiring manager, podcast host, or potential collaborator. Use whenever the user wants to reach out to someone they don't know personally.
Refine a rough SPEC.md into an implementation-ready specification through a multi-round interview, then break it into ordered tasks. Use when starting a new feature or turning a rough idea into something a developer can build without asking questions.
| name | idea-storm-x4 |
| description | Generate a large idea set with four parallel agents, then filter it down through two debate rounds. Use when the user wants to brainstorm or explore possibilities for a project, product, feature, or strategy. |
A wide-then-narrow brainstorm on the user's topic. Breadth comes from agents that can't see each other's output; quality comes from making the later rounds argue for what survives.
The topic is in $ARGUMENTS. If it's empty or too thin to brief an agent with, ask the user what they're brainstorming about before launching anything.
Launch four agents in parallel, each working a different angle so their idea spaces barely overlap. For consumer product work the seams that divide cleanly are growth and virality, monetization, community and UX, and content and data. Pick the angles that actually fit this topic — a dev tool, an internal process, and a marketplace each split differently.
Give every agent the same context about the user's project and ask each for 50 numbered ideas, a short title and a line or two apiece. The volume target is real, not decorative: the tail of a long list is where the non-obvious ideas live.
Launch two agents in parallel, each taking 100 ideas (agents 1+2 for one, 3+4 for the other) and arguing its set down to the strongest 50, merging near-duplicates along the way.
Judge on feasibility, real impact, originality, taste, and fit for the audience. An agent that deletes the bottom half without weighing these hasn't done the job.
One agent takes the 100 semifinalists, merges what overlaps across the two branches, applies the same criteria, and returns a final 50.
Show the final list, numbered and clean. Then ask which ones the user wants to pursue or turn into tasks.
