Markup AI Content Guardian

June 3, 2026

Version updated for https://github.com/markupai/content-guardian-action to version v2.0.0.

This action is used across all versions by 1 repositories.

Action Type

This is a Node action using Node version 24.

Go to the GitHub Marketplace to find the latest changes.

What’s Changed

⚠️ v2 is a breaking release — migration guide

v2 talks to the Markup AI agentic API directly and runs the style agent per file. Most workflows migrate in two edits: bump the version tag and replace the old style inputs (dialect, tone, style-guide) with the single optional style_guide.

Input changes at

v1 input	v2	What to do
`markup_ai_api_key`	Unchanged	Keep as-is
`github_token`	Unchanged	Keep as-is
`dialect` (required)	Removed	Configure on your style guide at console.markup.ai
`tone` (optional)	Removed	Configure on your style guide at console.markup.ai
`style-guide` (e.g. `ap`)	Replaced by `style_guide`	Underscore name and new values — see below
`add_commit_status`	Unchanged	Keep as-is
`add_review_comments`	Unchanged	Keep as-is
`strict_mode`	Unchanged	Keep as-is
—	New: `paths`	Optional path whitelist
—	New: `dry_run`	Optional — run analysis but skip all GitHub writes

Outputs (event-type, files-analyzed, results) keep their names; the results JSON shape changed (see note at bottom).

1. Bump the version

- uses: markupai/content-guardian-action@v1
+ uses: markupai/content-guardian-action@v2

2. Replace the style inputs

with:
  markup_ai_api_key: ${{ secrets.MARKUP_AI_API_KEY }}
  github_token: ${{ secrets.GITHUB_TOKEN }}
  dialect: american_english
  tone: formal
  style-guide: ap

After (v2):

with:
  markup_ai_api_key: ${{ secrets.MARKUP_AI_API_KEY }}
  github_token: ${{ secrets.GITHUB_TOKEN }}
  # dialect / tone / style guide rules now live on the style guide itself
  # at console.markup.ai. Omit style_guide to use the org default, or pin one:
  style_guide: "AP Style" # target ID or display name (case-insensitive); optional

⚠️ A find-and-replace of style-guide → style_guide is not enough. The name changed from hyphen to underscore, and the accepted values changed: v1 took canonical names like ap/chicago; v2 takes a style guide (target) ID or display name from your org. Passing ap will fail unless you have a style guide named “ap”.

style_guide is optional — set a default style guide in the console and omit it entirely. List valid values with:

curl -H "Authorization: Bearer $MARKUP_AI_API_KEY" \
  https://api.markup.ai/style-agent/targets | jq '.[] | {id, display_name, is_default}'

3. Expect risk-based scoring by default

Risk labels are always shown — 🔴 High / 🟡 Medium / 🟢 Low / ✅ No issues, plus severity counts (H:_ M:_ L:_).
Numeric 0–100 scores appear only if your org has style_agent_numeric_scoring enabled. No fabricated scores otherwise.
If you gate branch protection on the commit status, its state is now derived from risk: error for High, failure for Medium, else success.

Note for downstream consumers

outputs.results is still a JSON string but the per-file object shape changed. scores is null when numeric scoring is off, so guard for null before reading scores.*.