Skip to main content

Review Results

After a run completes, open it from the Analysis History table (click Details) to review the verdict.

Run overview​

The run summary shows the Overall score together with run metadata: Model, Commit, Branch, Execution Time, and Total Tokens. When enabled, an Executive Summary and Key Policy Insights give you a high-level read before you dive into individual controls.

The Analysis History table lists each run with its Run ID, Git Ref, Execution Time, Status, Controls, and Score.

Per-control findings​

The controls table shows each Control, its Status, Metrics, and Score. Status values include Pass, Fail, Pending, Executing, Blocked, and Error. Use the search box ("Search by ID or title…") or Filter Passed to focus on what needs attention.

Open a control's Details to see its Control Findings, grouped by severity:

  • Critical Findings
  • Potential Findings
  • Informational Observations

The Evaluation Deep-Dive & Evidence section traces each verdict back to the concrete evidence — the files and line references the sub-agent used to reach its decision.

Working with a finding​

Each finding carries three actions:

  • Copy its location — a copy button beside the location puts path:line on your clipboard, ready to paste into your editor. A finding that reports something as missing copies just the path, since there is no line to jump to.
  • Copy fix prompt — puts a complete, self-contained remediation request on your clipboard: the control, the standard, the severity, the location and the cited code. Paste it into an AI assistant without having to explain the context first.
  • Re-scan this control — re-evaluates that one control against your current code, so you can check a fix without re-running the whole analysis. It always performs a genuine re-evaluation, even when Skip Existing (Analysis) is on for the project, and it creates a new run rather than overwriting the one you are looking at — the earlier result stays on record.

Finding descriptions are rendered as formatted text, so code references, lists and emphasis appear the way the analysis wrote them.

Scores​

Scores are on a 0–100 scale. The app shows them as a percentage and the regulatory report writes them as /100. A control that carries no score — not applicable, disputed, or never graded — shows a dash rather than a zero, so an ungraded control is never mistaken for a failing one.

Export​

From a run you can Export PDF for a shareable report, or open View Deep Dive & Evidence for the full per-control breakdown.

Discussing and overriding a result​

Controls and individual findings carry a comment thread and an attributed change history, so a disputed finding can be talked through in place and a status override is always traceable to whoever made it. See Comments & Change History.

If a result is wrong in a way you would like fixed rather than merely recorded, use Share Feedback.

Iterating

Fix the flagged findings in your code, then start another run with Run New Analysis. Enable Skip Existing (Analysis) to resume quickly by re-evaluating only what changed. See Run an Analysis.