Content governance

The same mission and verify graph that governs TypeScript imports governs marketing copy. The content domain scans .md, .html, .txt and .json files and enforces pattern rules: forbid_pattern for claims that must never appear, require_pattern for boilerplate that must.

Walk through the example

The repository ships a fixture in examples/content-governance/ with three seeded violations:

The fixture ships its TARGET_ARCHITECTURE.yaml. gantry judges the repository it runs in, so copy the fixture into its own repository first:

cp -r examples/content-governance /tmp/content-governance && cd /tmp/content-governance
git init -q && gantry init --yes --no-hooks --no-ci   # keeps the fixture's TARGET_ARCHITECTURE.yaml
gantry arch check
# expect the seeded violations; fix the files and rerun until it passes

Do it in your repository

Write TARGET_ARCHITECTURE.yaml at the repository root with domain: content and one rule per claim or required block, then run gantry arch check. Each rule is a plain regular expression, so a rule matches or it does not, and reruns on an unchanged tree are byte-identical:

schema_version: "0.3.0"
domain: content
scan_roots:
  - content
layers:
  - id: content
    globs:
      - content/**
rules:
  - id: forbid-claim
    from_layer: content
    applies_to: ["content/**"]
    forbid_pattern: "(?i)cures cancer"
  - id: require-disclaimer
    from_layer: content
    applies_to: ["content/**"]
    require_pattern: "These statements have not been evaluated by the FDA"

In a mission

Use gantry arch check as the mission gate, or as one step of it. A violation comes back as a finding with offending_file and line, so a writing agent can fix the exact sentence:

{
  "failed_gate": "arch",
  "offending_file": "content/ad-bad.md",
  "line": 3,
  "severity": "error",
  "resolution_hint": "..."
}

Corpora without a built-in domain, such as accounting or legal text, use gate_command scripts and edit-root globs in the same graph (Perimeters).