How ratings are assessed

Method and limits

Ratings 2026-10

Evidence confidence

A rating rests on primary sources when at least one linked source is vendor documentation, an official repository or a changelog; vendor articles, community sources, vendor pages and research are secondary for this count. The product line counts ratings with at least one primary source and separately counts ratings marked unverified (?).

Coverage levels

Coverage levels are ordinal descriptions of documented support. They are not percentages and do not measure how much work is automated.

• 0 — Not supported: used only when a documented restriction supports that conclusion.
• 1 — Limited
• 2 — Substantial
• 3 — Complete
• ? — Unverified: evidence is insufficient or unclear; a gap in evidence is not a zero.

Each criterion page defines the evidence expected at levels 1, 2 and 3.

Whether a check starts automatically and whether a failure blocks delivery are described separately from the coverage level.

Rating rules

• Ratings describe documented product support. Missing or uncertain evidence is not treated as a negative claim.
• Structural freshness checks, executable behaviour tests, AI review and human decisions are distinct mechanisms.
• Generic executable quality checks without requirement mapping do not establish requirement-linked test coverage. Generic Git history does not establish a delivery audit trail.
• A gate that shows only that tests exist does not show that they passed.
• Governance is discussed in the report text, not assigned a coverage level.
• Each rating names the assessed product bundle and settings.

Limits

These are documentation assessments. No product trials were run.

The source-link check covered 303 destinations as an availability snapshot from October 2026. It did not re-verify the claims.