Sample Council Case: Selecting a Foundation Content Architecture

Synthetic sample — not an observed Council run

This record is deliberately synthetic. It demonstrates the minimum public Council format without claiming that named commercial models performed the roles, without exposing private prompts, and without presenting fictional execution as evidence.

Mission

Select a content architecture for the foundation. The two candidate directions are repository-managed MDX and an early database or headless CMS.

The acceptance criteria are versioned content, validated metadata, deterministic builds, stable routes, a reviewable public history, no runtime secrets, no arbitrary remote content execution, and low foundation complexity.

Lead proposal

The architecture lead proposes repository-managed MDX. Content lives beside the code, frontmatter is validated with Zod, visible drafts are explicitly marked, and the build can generate routes and search documents without a database.

The lead argues that this matches the public repository's transparency goal and keeps the foundation static-first.

Challenger findings

The challenger notes real costs. Repository-managed content asks editors to use Git, merge conflicts can affect prose, and a later editorial team may want previews, workflow permissions, scheduling, and structured reuse beyond files.

An early CMS could improve nontechnical editing and workflow, but it would also add authentication, network dependency, data migration, runtime configuration, and another public-private boundary before demand is measured.

Verification

The verifier checks the proposal against locked decision SF-011 and the foundation acceptance criteria. Local MDX satisfies the selected architecture when the implementation also rejects path traversal, symlinks, frontmatter drift, imports, exports, JSX, raw HTML, and executable MDX expressions.

A database or CMS is not inherently wrong. It is unnecessary for the approved foundation.

Decision and change from challenge

Choose repository-managed MDX now. The challenge improves the decision by recording a revisit trigger: demonstrated editorial friction or a dynamic product need, supported by an approved architecture change.

Stable slugs, source metadata, correction history, publication state, and migration-friendly schemas are required so a later architecture can preserve provenance.

Limits

This sample did not benchmark an editorial team, test a CMS, measure migration effort, or conduct a real multi-model session. It teaches a decision record; it is not proof that MDX is universally superior.