# AI & Agentic Red Team

> Scoped per engagement · AI Security · https://cisomarketplace.services/services/ai-agentic-red-team

Offensive testing of LLMs and AI agents — prompt injection, jailbreaks, tool misuse, agent goal hijack, memory/context poisoning, model extraction, and agent identity abuse. Mapped to OWASP Top 10 for Agentic Applications, OWASP LLM Top 10, and MITRE ATLAS.

## In scope

- Prompt injection & jailbreak testing
- Agent tool misuse & goal hijack
- Memory/context poisoning
- Model extraction & inference attacks
- Agent identity & privilege abuse
- MCP/tool-connection abuse
- OWASP Agentic + MITRE ATLAS mapping

## Deliverables

- AI red team report
- OWASP/ATLAS-mapped findings
- Exploit chains & PoCs
- Guardrail & mitigation recommendations
- Re-test of fixes

## Tiers

### Essential: scoped

Adversarial testing of 1 AI application or agent.

Limits: ai systems 1

Includes:
- One AI application or agent
- Findings report
- Fix guidance

Excludes:
- Tool and connector integrations beyond the first

### Advanced: scoped

Up to 3 AI applications or agents, including their tool and data connections.

Limits: ai systems 3

Includes:
- Up to three systems
- Tool and data connections

Excludes:
- Ongoing testing

### Enterprise: scoped

Up to 8 systems, multi-agent workflows and a retest after fixes.

Limits: ai systems 8

Includes:
- Up to eight systems
- Multi-agent workflows
- Retest

Excludes:
- Continuous testing — scoped as a subscription

## Terms

Starting price shown is the lowest published tier. A written proposal fixes scope, tier and price before anything is signed. Timeline is set in the SOW.

Scope this engagement: https://cisomarketplace.services/scope?service=ai-agentic-red-team