# Review Designs With Specialized AI Agents

Single-agent designs can hide assumptions, constraints, and failure modes. This skill runs a gated review with specialized agents and records every decision.

## Install

```bash
npx skillstore add sickn33/multi-agent-brainstorming
```

## Metadata

- Status: approved
- Slug: sickn33-multi-agent-brainstorming
- Skillstore revision: r2
- Version status: missing
- Tree hash: 55cf11da400ef617c13f59786dd9aba5a10f99ae83def665abb6d3200ae7bec3
- Author: sickn33
- GitHub username: sickn33
- License: MIT
- Repository: https://github.com/sickn33/antigravity-awesome-skills/tree/main/skills/multi-agent-brainstorming
- Ref: 81e05e636292629114b76cbb3922fbe57672fc02
- Supported tools: Claude, Codex, Claude Code
- Audit status: complete
- Agent install advisory: allowed
- Manual install advisory: allowed
- Artifact signature: available
- Audit attestation: unavailable
- Human verification: not\_verified
- Risk factors: external\_commands
- Quality score: 78
- Quality tier: bronze
- Public page: https://skillstore.pages.dev/skills/sickn33-multi-agent-brainstorming
- Manifest: https://skillstore.pages.dev/api/skills/sickn33-multi-agent-brainstorming/manifest

## Capabilities

- Assigns distinct designer, skeptic, constraint guardian, user advocate, and arbiter roles.
- Runs reviewers sequentially to keep feedback scoped and prevent uncontrolled idea expansion.
- Surfaces assumptions, edge cases, usability concerns, and non-functional constraints.
- Requires the designer to answer objections and revise the design when needed.
- Produces a decision log with alternatives, objections, resolutions, and rationale.
- Ends with an explicit approved, revise, or reject disposition.

## Use Cases

- Validate a System Architecture: Review a proposed architecture for failure modes, operational constraints, user impact, and unresolved assumptions before implementation.
- Challenge a Product Workflow: Test a workflow against user expectations, confusing states, poor defaults, and stated product requirements.
- Prepare a Technical Decision: Create a traceable decision log that records alternatives, objections, resolutions, and the final disposition.

## Prompt Templates

### Review a Simple Design

```
Review this design using the multi-agent process: [design]. Confirm the goal, constraints, and success criteria before starting.
```

### Stress-Test an Architecture

```
Evaluate this architecture: [architecture]. Focus on reliability, security, scalability, maintainability, cost, and user impact. Record every objection and resolution.
```

### Resolve Competing Options

```
Compare these design options: [options]. Use the stated requirements: [requirements]. Challenge assumptions, resolve objections, and produce a justified final disposition.
```

### Run a Constrained Production Review

```
Review this production design: [design]. Enforce these limits: [constraints]. Assess failure scenarios, privacy, operations, usability, and cost. Return the decision log and disposition.
```

## Limitations

- The process simulates specialist review and does not replace qualified human experts.
- Review quality depends on clear requirements, constraints, and success criteria.
- The skill evaluates designs but does not implement or test them.
- It depends on a separate standard brainstorming skill for the initial design phase.

## Best Practices

- Provide measurable requirements, constraints, and success criteria before review begins.
- Keep each reviewer within its assigned role and address every objection explicitly.
- Preserve the decision log with the final design for later implementation and validation.

## Anti Patterns

- Do not start review with an undefined problem or missing constraints.
- Do not let reviewers add unrelated features or redesign outside their mandate.
- Do not treat simulated consensus as proof that the design works in production.

## Security Audit

- Audited at: 2026-08-04T15:08:00.574\+00:00
- Summary: All four static findings are false positives caused by Markdown backticks or ordinary review language. The skill contains no shell execution, system reconnaissance, prompt injection, or other semantic security issue.

## Stats

- Views: 123
- Downloads: 11
- Favorites: 0
- Popularity score: 0