Logic Guard

Logic Guard reviews every artifact in a project against a catalog of governance rules and reports what it finds: broken transitions, outputs that are never set, unbounded queries, references to deleted columns, and many other issues that the designer doesn’t flag as errors. Each issue has a severity, an explanation and a suggested fix. Use it as a quality gate before you publish or promote a solution.

Logic Guard only reads your project. It never changes it.

Scan a project

1. Sign in to Logic Guard. On Scan, select the project.
2. Select Analyze Project. Results appear as each section is checked.
3. Open Results. Issues are grouped by section (artifact type) and by artifact, each marked Critical, Warning or Info. Most Problematic Sections shows where to start.
4. Expand an issue to read its Explanation and Suggested fix.
5. Select Download HTML to share the report.

Logic Guard reports counts per severity rather than a single score, so one critical issue is never hidden behind many minor ones.

Tune the rules

Open Rules to change how a project is checked. Only members of the project can change its rules.

Area	What you can do
Baseline Catalog	Every built-in rule is checked unless you disable it here. You can disable a rule or change its severity, and reset it to the catalog default.
Custom Rules	Select Add Rule to add your own check per artifact type: a rule ID (for example CUSTOM-WF-001), a short title, a description in plain language, a severity and a suggested fix.
Strictness	Conservative reports only high-confidence, severe issues. Standard is the default balance. Thorough surfaces every plausible issue.
Findings language	The language of explanations and fixes. English is the default.
Glossary	Maps terms to your own, so findings use your vocabulary.
Additional instructions	Guidance for every check, for example Treat all naming-convention findings as Warning for this project.
Section-specific context	A note for one section only, for example Flows prefixed TMP_ are scratch flows; don’t flag hardcoded thresholds in them.

Select Save Rules to keep rule changes. Scan settings shape how issues are reported; they don’t change which rules run.

How it behaves

• Checked against your live schema. References to deleted tables or columns are reported.
• Across artifacts. Issues that exist only between artifacts, such as a reference from one artifact to another that doesn’t match, are reported too.
• Runs as you. It reviews only what you can read.
• Private jobs. Only the person who started a scan can see and steer it while it runs.

Example

Before promoting the Service project to test, you run Logic Guard with Thorough strictness. It reports a critical issue in a business process (an unreachable stage) and several warnings in logic flows. You fix the stage, add a section-specific note that TMP_ flows are experiments, rescan, and attach the HTML report to the release.

Related

• Logic Assurance
• Logic Sentinel
• Composer best practices
• Process Orchestrator best practices