/gaia-adversarial
user-facingWhat it does
/gaia-adversarial performs a cynical, skeptical adversarial review to surface flaws, gaps, and weaknesses in any document, design, or code. It assumes nothing works as claimed and attacks from multiple angles -- technical, business, user, security, and scale. Findings include severity and confidence levels but no fixes -- identifying problems is the sole purpose.
When to use it
- You have a PRD, architecture document, or design that has not been stress-tested and you want a ruthless critique before committing to implementation.
- You want to find the weaknesses in your plan -- gaps, unstated assumptions, scaling risks, security holes -- before your users or competitors find them.
- You need a second opinion that deliberately assumes the worst case.
Prerequisites
- No strict prerequisites. Works best when planning artifacts exist for context.
How to invoke
/gaia-adversarial .gaia/artifacts/planning-artifacts/prd.md
What it does step by step
- Load target Reads the target document, design, or code.
- Adversarial analysis Attacks from ten perspectives: feasibility, completeness, contradictions, assumptions, scale, failure modes, dependencies, security, user impact, and business risk.
- Generate report Produces a ranked findings report with severity and confidence levels. Writes to
.gaia/artifacts/creative-artifacts/adversarial-review-{target}-{date}.md.
Inputs
| Input | Source | Description |
|---|---|---|
topic | Command argument (optional) | The topic or challenge. If omitted, the command asks. |
Outputs
| Output | Location | Description |
|---|---|---|
| Creative artifact | .gaia/artifacts/creative-artifacts/ | The session output document. |
Example session
> /gaia-adversarial .gaia/artifacts/planning-artifacts/prd.md
10 findings:
| Severity | Confidence | Finding |
|----------|------------|--------|
| critical | high | Auth token refresh not specified -- session loss risk |
| high | medium | Scale: no pagination on user listing |
| medium | high | Assumption: third-party API SLA not verified |
Written to .gaia/artifacts/creative-artifacts/adversarial-review-prd-2026-05-07.md
What to run next
- Address the highest-severity findings, then run the adversarial review again to check your fixes.
- Feed findings into
/gaia-add-featureor/gaia-edit-prdto update the affected artifacts.
Troubleshooting
Topic not provided
The command asks for your topic interactively if no argument is given.