Readability Profiles

The TameAGENTS readability coach analyzes every agents.md and highlights structural gaps before rules reach your agents. Use it to quickly spot missing metadata, inconsistent headings, or thin rule sets.

TameAGENTS readability demo

Scoring rubric

Each agents.md receives a composite score based on four pillars. The sidebar badge in VS Code mirrors this rubric so authors know exactly why a document was flagged.

  • Frontmatter completeness: profile, leash settings, imports, metadata
  • Structural coverage: presence of overview, rules, and escalation steps
  • Actionability: minimum rule count, directives per section
  • Link density: pointers to owners, docs, or repos for deeper guidance

Live diagnostics

The readability coach streams inline diagnostics as you type. Hover over the margin badge to see which rule triggered and how to fix it. Suggestions cover everything from missing headings to low signal-to-noise ratio.

Workspace dashboards

Open the dashboard (⌘⇧R / Ctrl+Shift+R) to see aggregate scores, filter by team, and export a CSV of hotspots. Every row links back to the responsible repository path.

Configuration

Fine-tune thresholds in tameagents.readability.* settings:

{
  "tameagents.readability.minRuleCount": 4,
  "tameagents.readability.requireOwners": true,
  "tameagents.readability.checklists": [
    "All sections include at least one actionable directive"
  ]
}

Tip: bake it into CI

Combine the readability coach with tameagents report conflicts --json to fail builds when new folders regress. TameAGENTS surfaces both lint and inheritance conflicts in one pass.