01
MCP vs Agent Skills: Reach, Know-How, and the Controls Between Them
A practical guide to deciding when an AI system needs MCP, a Codex Skill, or both—and how to keep the resulting workflow accountable.
8 minPractical field guide
Field notes
Reach, know-how, and value
Agent Systems and AI Economics
Architecture and measurement patterns for giving AI systems useful reach, reliable know-how, and accountable business outcomes.
- Built for
- CTOs, engineering leaders, operations leaders, and AI program owners
- Path length
- 3 chapters · 26 minutes
02
Secure MCP Integration Patterns for Customer-Facing AI
A practical architecture guide for MCP identity, least privilege, approvals, audit trails, retries, and failure containment.
9 minPractical field guide
03
How to Measure AI Contribution Without Pretending Correlation Is ROI
A practical measurement model for connecting AI activity and cost to work delivered while keeping attribution, causation, and ROI distinct.
9 minPractical field guide