Overview:

In order to properly start with ACS MCP development we need to define correct requirements.

To gather functional requirements - current plan is:

• collect list of use cases for ACS MCP from different sources (consultants, field people, developers, internal customers, etc.)
• define AI Agent prompts that would fulfill list of use cases
• identify required information from internal and external sources
• categorize prompts by complexity and provided value
• define required information that should be fetched from ACS to fulfill defined prompts

Non-function requirements:

• align with Red Hat MCP best practices
• use MCP libraries with stable releases
• integrate MCP evaluation with at least 2 models
• create a separate project from stackrox with different release cadence and CI pipeline

UPDATE

Relevant links:

• Prompts with complexity and required ACS data: https://docs.google.com/spreadsheets/d/1F3VHefjtMQ07vDaHms1n_t36dEBcBC7FnqvyHy0Rrhg/edit?gid=0#gid=0
• V1 doc with use cases: https://docs.google.com/document/d/1toa3JllvJAcDLYhd_GotnMTpnlEMZkLQcZ8V5NBTbt8/edit?tab=t.0#heading=h.4ncoor7tyxh3
• Gathered feedback: https://docs.google.com/document/d/1AOZUQr6y2euTGDjF-3eB7k1m2_j8NvwSbxSMlDrHhtM/edit?tab=t.lilb95p8vhix