Run the first analysis

The first analysis builds the index of a project. It parses every source file, so it takes longer than later analyses, which only re-parse changed files. This topic covers the first run. Keep the index current covers everything after it.

Prerequisite: the exploration is open and shows Review before analyzing. A new exploration always opens there.

Decide what to analyze

The graph and the AI's answers are only as good as the files in the index. Generated code, vendored libraries and test fixtures add many symbols that nobody works on and push the relevant ones out of search results and hotspot lists.

1. On Review before analyzing, read Suggested exclusions. Tracemoor suggests test, spec and story files and folders such as tests, __mocks__, examples and fixtures.
2. Tick the suggestions that do not belong in the graph. Keep the tests if you want to see which code they call.
3. Click Analyze.

Review before analyzing lists the folders the analysis always skips, such as node_modules, dist and target, and the files that .gitignore excludes.

Exclude more after the first run

Folders that the suggestions do not cover show up after the first run, for example a generated/ folder at the top of Complexity Hotspots or a copy of a library under third_party/.

1. In the file tree, right-click the folder and click Exclude this folder.
2. Choose Analysis › Re-analyze Project.

See Exclude files and folders. Exclusions are stored per machine and are not shared.

Check the project profile

During the analysis Tracemoor reads the manifest files of the project (package.json, Cargo.toml, pyproject.toml, go.mod, pom.xml and others) and counts files per language. The result is the project profile.

To see it, choose View › Source View. While no symbol is selected, the Details panel on the right shows the profile under Project.

A Tauri app with a React frontend. src-tauri is detected as the backend of the Tauri app.

For a VS Code extension the panel also shows the contributions of each extension package. Open Manifest → or View › Extension Manifest opens the full manifest.

The profile describes the project. It does not change how files are parsed. One part of it has an effect: for a project detected as a VS Code extension, Analysis › Connect Browser… starts the VS Code inspector instead of the web inspector. If the project also has a web frontend, it asks which one.

Correct the profile

If the profile is wrong, add the file .codeexplorer/project_overrides.json to the project (not to the exploration) and re-analyze. The Details panel then shows override active.

{
  "project_kinds": ["desktop_app", "multi_language"],
  "tech_items": {
    "add": [{ "category": "framework", "name": "SwiftUI", "sub_root": "gui" }],
    "remove": [{ "category": "framework", "name": "AppKit" }]
  }
}

• project_kinds replaces the detected kinds. Allowed values are listed in Project kinds.
• tech_items.add and tech_items.remove change single entries. category is one of framework, language, runtime, bundler, database, test, other. sub_root names a sub-project folder; without it the entry belongs to the project root.

A file with an unknown value is ignored as a whole.

Set the cluster root

The analysis groups files into clusters. The direct subfolders of the cluster root become the top-level clusters. Tracemoor picks src, then lib, then app, then the project root. The clusters are used by the AI's structure tools and by the coverage drill-down.

1. Choose Analysis › Project Settings….
2. On the General tab, enter a folder under Explicit Cluster Root (optional) or click Browse…. Reset to Auto returns to the detected folder.
3. Save and re-analyze the project.

Example: in a monorepo whose code lives under packages/, set the cluster root to packages so that each package becomes a cluster.

Next step

Continue with Find your way in the codebase.