[PRODUCT] Remote MCP strategy decision — INTERNAL

[LucielAI]

Classification: Internal Product Decision

Do NOT activate implementation without explicit approval.

Context

stdio-only MCP is now the biggest strategic gap if SPINE ever wants to support:

• Remote/cloud-hosted agents
• Multi-agent coordination scenarios
• Agents running on different machines from SPINE's governance state

The external agent review (2026-04-10) identified this as the most significant strategic question post-beta exit.

Strategic question

Should SPINE open a remote MCP path at all, or deepen the local-first wedge?

Option A — Stay local-first:

• Strengthen the current local wedge (SPINE_ROOT ergonomics, discipline-tax reduction, OpenClaw first-class)
• Remote MCP is a Phase 4+ question
• Keeps SPINE simple, inspectable, and without network attack surface

Option B — Open remote MCP as Phase 3B:

• Adds HTTP/MCP transport as an optional SPINE capability
• Enables cloud-hosted agent scenarios
• Increases complexity and operational surface

Decision criteria

Before choosing Option B, validate:

• Is there a concrete, funded use case for remote MCP?
• Does the complexity increase break SPINE's local-first trust model?
• Would it turn SPINE into a platform instead of a CLI?

Triage classification

DEFER UNTIL EXPLICIT APPROVAL

This is a product strategy decision, not an implementation task. Do not create issues or start work until Hashi explicitly approves one of the options above.

---

Internal — deferred until explicit product decision