Migration Readiness

Migration readiness is a 0–100 score that answers one question: what fraction of your detected direct LaunchDarkly calls can FlagLint safely automate?

Formula

safely automatable calls ÷ total detected direct LaunchDarkly calls × 100

A call is counted as safely automatable when FlagLint can statically prove that the flag key is a string literal, the fallback value and type are known, the LaunchDarkly source client is traceable to a supported SDK import, and a configured OpenFeature destination binding is present. Every other call — dynamic keys, detail evaluations, bulk calls, ambiguous clients, unknown fallbacks — is counted as requiring manual review and is excluded from the automatable numerator.

Grade Thresholds

Score	Grade	Meaning
≥ 80	ready	Most calls are safe to automate. Migration can proceed with minimal manual effort.
50–79	moderate	A significant portion of calls require manual review. Plan time for manual migration work.
< 50	complex	More than half of calls require manual review. Prioritize resolving manual-review signals before migrating.
—	not-applicable	No direct LaunchDarkly calls detected. Score is null and no readiness bar is displayed.

Manual-Review Breakdown

The following categories explain why a call is counted as requiring manual review. This breakdown is informational — it helps you understand where effort will be needed, but it does not numerically alter the score beyond those calls being counted as non-automatable.

Manual-review reason	Why review is required	Recommended next step
Dynamic flag key	Flag key determined at runtime — cannot be statically rewritten	Extract key to a named constant
Detail evaluation	boolVariationDetail/variationDetail — returns metadata with no direct OpenFeature equivalent	Migrate manually or check OpenFeature provider metadata support
Bulk evaluation	allFlagsState — no single-flag codemod exists	Requires architecture decision; migrate flag by flag
Unknown fallback value	Fallback type cannot be inferred statically	Add an explicit typed fallback literal
Ambiguous client binding	Call site cannot be proven to use a supported LD client	Ensure client provenance is traceable from a supported SDK import

Not-Applicable Behavior

When no direct LaunchDarkly calls are detected, the readiness score is null and the grade is not-applicable. This happens when:

• The scanned directory contains no LaunchDarkly SDK imports.
• All detected calls are from unsupported SDKs (browser, React, non-Node).
• The directory is empty or contains only test files excluded via --exclude-tests.

No readiness bar or score line is printed to the terminal in this case.

Further Reading

• flaglint audit CLI reference — command options and example output
• Cost Estimation (—cost-estimate) — directional effort planning