brainstorming-extension

Enforces a five-agent design review gate after brainstorming before implementation planning begins.

1|Updated Jun 23, 2026
One-click install
npx skills add https://github.com/seal-harness/seal-harness --skill brainstorming-extension-seal-harness
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: brainstorming-extension
Source: https://github.com/seal-harness/seal-harness/tree/main/.agents/skills/brainstorming-extension
Command: npx skills add https://github.com/seal-harness/seal-harness --skill brainstorming-extension-seal-harness

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? The upstream brainstorming workflow commits a design document and jumps directly into implementation planning, skipping any structured review. This Skill closes that gap by inserting a mandatory design review gate so architectural, security, and requirements issues are caught before expensive implementation begins. ## Core Features & Use Cases - Mandatory Review Gate: Routes every committed design document through five parallel review agents (Product Manager, Architect, Designer, Security Design, CTO) that must all approve before planning proceeds. - Redundant Enforcement: Uses three mechanisms — CLAUDE.md workflow rules, the /start-task command handoff block, and this skill document — so the gate cannot be silently bypassed. - Iteration and Escalation: Supports up to three revision cycles when reviewers request changes, then escalates to a human decision (Override, Defer, or Cancel). - Use Case: After brainstorming a new feature design, the Skill automatically spawns the five-agent review panel, reports a verdict table, and only then allows writing-plans and orchestrated execution to proceed. ## Quick Start After brainstorming finishes and the design document is committed, run the design review gate on it and wait for all five reviewers to approve before starting implementation planning.

Frequently Asked Questions about brainstorming-extension

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I add a design review step after brainstorming?

Invoke the design review gate immediately after the brainstorming skill commits a design document. This spawns five parallel review agents — Product Manager, Architect, Designer, Security Design, and CTO — and blocks planning until all five approve.

What happens if a design review agent rejects the design?

If any reviewer returns NEEDS_REVISION, the blocking issues and open questions are reported back, and you revise the design document before re-running the gate. After three failed iterations, the decision escalates to a human who can override, defer, or cancel.

Can I skip the design review gate for small changes?

The gate can be skipped only when you explicitly request it, the agent confirms the skip with you, and the design is genuinely simple — under one day of work and fewer than three files. Otherwise the review is mandatory.

How is the review gate enforced if triggers are only metadata hints?

Enforcement relies on three redundant mechanisms: a mandatory workflow section in the CLAUDE.md template loaded into every conversation, a handoff block in the /start-task command, and this skill document as the detailed reference procedure.

What comes after the design review gate passes?

Once all five reviewers approve, the workflow proceeds to creating an epic, writing an implementation plan, running that plan through a separate plan review gate with three adversarial reviewers, and then starting orchestrated execution in an isolated worktree.