Struct AnalysisOutput 

pub struct AnalysisOutput {
    pub results: AnalysisResults,
    pub timings: Option<PipelineTimings>,
    pub graph: Option<ModuleGraph>,
    pub modules: Option<Vec<ModuleInfo>>,
    pub files: Option<Vec<DiscoveredFile>>,
    pub script_used_packages: FxHashSet<String>,
    pub file_hashes: FxHashMap<PathBuf, u64>,
}

Result of the full analysis pipeline, including optional performance timings.

Fields

results: AnalysisResults

timings: Option<PipelineTimings>

graph: Option<ModuleGraph>

modules: Option<Vec<ModuleInfo>>

Parsed modules from the pipeline, available when retain_modules is true. Used by the combined command to share a single parse across dead-code and health.

files: Option<Vec<DiscoveredFile>>

Discovered files from the pipeline, available when retain_modules is true.

script_used_packages: FxHashSet<String>

Package names invoked from package.json scripts and CI configs, mirroring what the unused-deps detector consults. Populated for every pipeline run; trace tooling reads it so trace_dependency agrees with unused-deps on “used vs unused” instead of returning false-negatives for script-only deps.

file_hashes: FxHashMap<PathBuf, u64>

xxh3 content hash of every parsed source file, keyed by absolute path. Used by fallow fix to detect on-disk drift between the in-process analysis read and the per-file write; if the file’s current hash differs from the captured value, the fix for that file is skipped with a clear diagnostic and exit 2. The hash is the same value extract/cache uses for cache invalidation, so a cached parse contributes the same hash as a fresh parse. Roughly 8 bytes per file (negligible memory cost even on 100k-file projects).