| name | vams-red-team |
| description | Perform adversarial VAMS security review for protocol designs, diffs, PRs, deployments, gateway routes, smart contracts, DA adapters, bridges, session keys, TEE trust, OMS identity, token economics, and cognitive-layer memory/search systems. Use when the user asks for red team, attack analysis, abuse cases, exploit paths, or pre-audit critique. |
VAMS Red Team
Use this skill to think like a capable adversary while staying within defensive
review boundaries. Do not provide exploit code for live systems.
Workflow
- Define assets, trust boundaries, attacker roles, and deployment stage.
- Read relevant code and docs; do not rely on architecture claims.
- Use
references/threat-model.md to enumerate attack classes.
- Identify practical exploit paths and required preconditions.
- Rank findings by exploitability, blast radius, and invariant impact.
- Provide concrete mitigations and tests.
- Block deployment if a high-impact path is unmitigated.
Output Shape
- Attack Surface: components and trust boundaries.
- Findings: severity, path, preconditions, impact, evidence.
- Invariant Impact: affected INV IDs.
- Mitigations: specific code/config/test changes.
- Residual Risk: what remains after fixes.
Keep the review technical and grounded. Avoid generic security checklists unless
they map to an actual VAMS path.