Source Write-back¶

The -apply-tags modifier instructs MethodAtlas to insert AI-generated display names and tags directly into the scanned Java (.java) and C# (.cs) source files, instead of writing a CSV report.

When to use this mode¶

You want to apply AI-suggested annotations to test source files automatically, without manually editing each file.
You have completed an AI enrichment run (via API AI enrichment or the Manual AI workflow) and want to persist the suggested labels as source annotations.
You want to annotate tests so that tag-based dashboards and CI gates reflect the AI's security classification immediately, without waiting for a separate annotation step.
You have an existing override file with reviewed classifications and want to replay those decisions as source annotations without re-running AI.

If you want human review of each annotation decision before any source file is touched, use Apply Tags from CSV instead — it separates the review step from the write-back.

It can be combined with API AI enrichment or with the Manual AI workflow consume phase.

With API AI enrichment¶

./methodatlas -ai -apply-tags \
  -ai-provider ollama \
  -ai-model qwen2.5-coder:7b \
  src/test/java

With Manual AI workflow¶

./methodatlas -manual-consume ./work ./responses \
  -apply-tags src/test/java

With an override file¶

When the codebase has been classified in a previous session and results are captured in an override file, apply-tags can use those results directly without re-running AI:

./methodatlas -override-file .methodatlas-overrides.yaml \
  -apply-tags src/test/java

This is the most efficient workflow after an initial classification: run AI once, review and commit the override file, then use it for subsequent annotation write-backs without additional AI calls.

What gets written¶

Only test methods classified as security-relevant receive new annotations. For each such method MethodAtlas inserts:

@DisplayName("...") with the AI-suggested display name as the value.
One or more @Tag("...") annotations for each taxonomy tag returned by the AI.

Annotations are inserted using JavaParser's lexical-preserving printer, which preserves the original formatting of all surrounding code, including whitespace, comments, and import ordering. Existing annotations on the same method are not modified or removed.

Summary output¶

After the run, a summary line is printed to standard output:

Apply-tags complete: 12 annotation(s) added to 3 file(s)

No CSV is produced.

Modifies source files in place

-apply-tags edits .java files directly. There is no dry-run mode. Commit or back up your work before running.

Reviewing changes¶

After write-back, inspect the diff before committing:

git diff src/test/java

AI-generated annotations are a starting point. Review the suggested @DisplayName values and @Tag assignments for accuracy before merging them into the main branch.

End-to-end scenario: annotating a legacy test suite¶

A team has a large legacy test suite with no @Tag or @DisplayName annotations. They want to quickly annotate all security-relevant tests so that their CI security gate can filter on @Tag("security").

# Step 1: run AI enrichment with write-back enabled
./methodatlas \
  -ai \
  -ai-provider ollama \
  -ai-model qwen2.5-coder:7b \
  -apply-tags \
  src/test/java

Output:

Apply-tags complete: 23 annotation(s) added to 7 file(s)

# Step 2: review the changes
git diff src/test/java

A typical diff looks like:

+import org.junit.jupiter.api.DisplayName;
+import org.junit.jupiter.api.Tag;
+
 @Test
+@Tag("security")
+@Tag("auth")
+@DisplayName("SECURITY: auth — login is rejected when the token has expired")
 void loginWithExpiredToken() {

# Step 3: commit reviewed annotations
git add src/test/java
git commit -m "chore: apply MethodAtlas security annotations to legacy tests"

After this commit, any CI pipeline that filters on @Tag("security") or uses MethodAtlas with -security-only will include these tests in its security inventory.

See CLI reference — -apply-tags for annotation placement guarantees and formatting details.