Sandbox-to-production preflight

Compare Jira automation rules across sites.

Export automation rules—now called automation flows in current Atlassian documentation—from sandbox and production. RuleSignal shows what changed, appeared, disappeared, or stayed the same while both JSON files remain in the browser tab.

2 filesbaseline and current export
4 stateschanged, added, removed, unchanged
0 uploadsfiles stay in the tab
0 scopesno Jira API access
Admin checklist

Review the change before import or activation.

01 · EXPORT

Export both environments

Use Jira's global automation export in the source and target sites. Keep the files below Jira's documented 5 MB import limit.

02 · IDENTIFY

Match rules conservatively

Prefer stable rule IDs. When site migrations change IDs, match only a name that is unique in both exports; never guess across duplicate names.

03 · DIFF

Separate four outcomes

Review changed, added, removed, and unchanged rules independently. A missing rule can matter as much as an edited one.

04 · LIMITS

Check structural pressure

Look for flows at the documented 65-step boundary, broad scheduled searches, deep branches, fan-out, and duplicate execution paths.

05 · GUARDS

Review risky actions

Confirm web requests, destructive actions, and self-triggering field updates have narrow conditions and idempotency protections.

06 · VALIDATE

Test live behavior separately

Static JSON cannot prove permissions, mappings, endpoint behavior, or execution history. Validate in a safe project before enabling imported flows.

Why a text diff is not enough

Site-specific IDs and JSON order create noise.

RuleSignal canonicalizes JSON for local SHA-256 content fingerprints, matches stable IDs first, and uses normalized names only when each name is unique on both sides. It refuses ambiguous duplicate-name matches and tells the reviewer when a heuristic was used.

A fingerprint identifies the reviewed content. It is not a digital signature and does not prove who created or approved the export.
Current Atlassian constraints

Ground the review in platform evidence.

Atlassian documents a normal-flow limit of 65 steps, a 999-item scheduled search limit, a 5,000-associated-item limit, and loop detection after rapid repeated triggering. Atlassian also advises narrowing JQL to the smallest relevant set. RuleSignal surfaces exported evidence related to these risks but does not replace runtime testing.

Honest product boundary

A deterministic preflight, not a runtime oracle.

RuleSignal uses 30 explainable checks and never sends the selected exports to a vendor backend or AI model. Missing evidence is marked not evaluated, never silently treated as a pass.