# $500 Focused AI Agent / MCP Security Assessment — Deliverable Specification

- **Offer URL:** `https://cainstudio.online/assessment/`
- **Turnaround:** 3 business days | **Price:** $500 fixed scope (no platform lock-in)
- **Customer Inputs Needed:** Tool JSON schemas / MCP server definitions, policy rules, or a staging endpoint.
- **Concrete Deliverables:**
  1. Attack-surface inventory mapped to OWASP Agentic Top 10 (ASI01–ASI10) and MCP threat vectors (tool poisoning, shadowing, replay, parameter mutation, cross-session memory poisoning).
  2. Executable test harness demonstrating fail-closed policy and capability-token behavior on the buyer's top 5 high-risk tool boundaries.
  3. Signed, hash-chained evidence bundle + standalone Python stdlib offline verifier (`0` vendor code required to re-check findings).
  4. Prioritized remediation roadmap and go/no-go recommendation for a $1,500 governance review or $5,000 MCPGate pilot.
