tobi/qmd

9.2

Adjusted Score

9.2

Raw Score

100%

Time Factor

2026-08-18

Last Push

29.3K

Stars

TypeScript

Language

59.2K

Lines of Code

178

Files

350

Pattern Hits

2026-08-29

Scan Date

0.01

HC Hit Rate

What These Metrics Mean

Adjusted Score: Primary synthetic code indicator. Raw score normalised per 1,000 lines of code and multiplied by the temporal discount factor. This is the definitive comparative metric — use it to rank repositories by AI authorship density.
Raw Score: The unmodified sum of all severity-weighted, context-multiplied pattern match scores before temporal discounting. Reflects the absolute signal strength independent of when the repository was last active.
Time Factor: The temporal discount multiplier (0–100%) applied to the raw score. Repositories last updated before ChatGPT's launch (Nov 2022) receive a 5% factor. Full signal is only assigned to repositories active in the post-adoption era (Jan 2024+).
Pattern Hits: Total count of individual pattern matches across all files and categories. A high hit count with a low score may indicate a very large codebase with isolated AI snippets; a low count with a high score indicates dense, concentrated AI signatures.
HC Hit Rate: High+Critical pattern hits per file, averaged across the repository. This orthogonal signal catches repositories where a few files are densely packed with high-severity AI tells — a strong indicator even when the normalised score appears moderate due to codebase size.
Lines of Code / Files: Total lines and files analysed. The scanner examines 94 file extensions. These denominators are used to normalise the score, enabling fair comparison between repositories of vastly different sizes.

Score History

This chart maps the temporal evolution of the adjusted synthetic code score across successive scan runs. An upward trajectory indicates ongoing incorporation of AI-generated code or expanding LLM-assisted scaffolding; a stable or declining trajectory may reflect active human refactoring, code removal, or the adoption of stricter authorship policies. The dashed secondary line (right axis) independently tracks total raw pattern hit count, which can diverge from the normalised score when codebase size changes significantly between scans.

Severity Breakdown

Classifies detected patterns by their diagnostic confidence and structural impact. CRITICAL patterns (coefficient 10) represent definitive synthetic signatures — hallucinated imports, explicit LLM attribution metadata — virtually never produced by human authors. HIGH (5) indicates strong structural tells such as cross-file repetition or cross-linguistic idioms. MEDIUM (2) covers recognisable conversational padding and AI-specific vocabulary. LOW (1) captures subtle indicators like tautological comments and generic boilerplate that require density to carry independent signal.

CRITICAL 0HIGH 1MEDIUM 142LOW 207

Directory Score Breakdown

This horizontal bar chart decomposes the repository's raw synthetic code score by top-level directory, allowing you to pinpoint precisely which modules or components carry the highest AI authorship density. Directories with disproportionately high scores relative to their size warrant targeted manual review: concentrated AI signatures often trace back to mass-generated configuration layers, auto-ported test suites, LLM-scaffolded boilerplate classes, or entire subsystems authored under heavy copilot assistance. Use this view to prioritise your human code-review effort.

Pattern Findings

The scanner identified 350 distinct pattern matches across 15 syntactic categories. Each entry below represents a discrete location in the source code where the engine recorded a statistically significant AI authorship indicator. Expand any category row to inspect the individual file paths, line numbers, code snippets, and the lexical context (CODE, COMMENT, or STRING) in which each match was detected.

Reading the findings table: The Severity column indicates the diagnostic confidence level (CRITICAL / HIGH / MEDIUM / LOW). The Context column identifies whether the match occurred inside executable code, an inline comment, or a string literal — comment-context matches receive a ×1.5 weight because LLMs systematically over-annotate. The ⚡ bolt icon marks clustered matches: three or more patterns within a 10-line window, each receiving an additional ×1.5 density multiplier as dense clusters constitute far stronger evidence of synthetic authorship than isolated hits.

Decorative Section Separators57 hits · 186 pts

Severity	File	Line	Snippet	Context
MEDIUM⚡	example-index.yml	12	# ─── Global context ──────────────────────────────────────────────────────────	COMMENT
MEDIUM⚡	example-index.yml	17	# ─── Terminal hyperlinks (optional) ──────────────────────────────────────────	COMMENT
MEDIUM⚡	example-index.yml	22	# ─── Model overrides (optional) ──────────────────────────────────────────────	COMMENT
MEDIUM⚡	example-index.yml	31	# ─── Collections ─────────────────────────────────────────────────────────────	COMMENT
MEDIUM⚡	example-index.yml	34	# ── Basic: just a path and a pattern ───────────────────────────────────────	COMMENT
MEDIUM⚡	example-index.yml	41	# ── Hierarchical context: path-prefix → description ────────────────────────	COMMENT
MEDIUM	example-index.yml	52	# ── Auto-update: run a command before re-indexing ──────────────────────────	COMMENT
MEDIUM	example-index.yml	63	# ── Ignore patterns: exclude files from indexing ───────────────────────────	COMMENT
MEDIUM	example-index.yml	79	# ── Exclude from default searches ──────────────────────────────────────────	COMMENT
MEDIUM	example-index.yml	89	# ── Everything together ────────────────────────────────────────────────────	COMMENT
MEDIUM	finetune/eval_retrieval.py	37	# =============================================================================	COMMENT
MEDIUM	finetune/eval_retrieval.py	39	# =============================================================================	COMMENT
MEDIUM	finetune/eval_retrieval.py	75	# =============================================================================	COMMENT
MEDIUM	finetune/eval_retrieval.py	77	# =============================================================================	COMMENT
MEDIUM	finetune/eval_retrieval.py	154	# =============================================================================	COMMENT
MEDIUM	finetune/eval_retrieval.py	156	# =============================================================================	COMMENT
MEDIUM	finetune/eval_retrieval.py	255	# =============================================================================	COMMENT
MEDIUM	finetune/eval_retrieval.py	257	# =============================================================================	COMMENT
MEDIUM	finetune/eval_retrieval.py	328	# =============================================================================	COMMENT
MEDIUM	finetune/eval_retrieval.py	330	# =============================================================================	COMMENT
MEDIUM	finetune/reward.py	25	# =============================================================================	COMMENT
MEDIUM	finetune/reward.py	27	# =============================================================================	COMMENT
MEDIUM	finetune/reward.py	86	# =============================================================================	COMMENT
MEDIUM	finetune/reward.py	88	# =============================================================================	COMMENT
MEDIUM	finetune/reward.py	136	# =============================================================================	COMMENT
MEDIUM	finetune/reward.py	138	# =============================================================================	COMMENT
MEDIUM	finetune/reward.py	263	# =============================================================================	COMMENT
MEDIUM	finetune/reward.py	265	# =============================================================================	COMMENT
MEDIUM⚡	finetune/reward.py	644	# =============================================================================	COMMENT
MEDIUM⚡	finetune/reward.py	646	# =============================================================================	COMMENT
MEDIUM	finetune/reward.py	662	# =============================================================================	COMMENT
MEDIUM	finetune/reward.py	664	# =============================================================================	COMMENT
MEDIUM	finetune/dataset/schema.py	31	# ---------------------------------------------------------------------------	COMMENT
MEDIUM	finetune/dataset/schema.py	33	# ---------------------------------------------------------------------------	COMMENT
MEDIUM	finetune/dataset/schema.py	78	# ---------------------------------------------------------------------------	COMMENT
MEDIUM	finetune/dataset/schema.py	80	# ---------------------------------------------------------------------------	COMMENT
MEDIUM	finetune/dataset/schema.py	114	# ---------------------------------------------------------------------------	COMMENT
MEDIUM	finetune/dataset/schema.py	116	# ---------------------------------------------------------------------------	COMMENT
MEDIUM⚡	finetune/dataset/schema.py	138	# ---------------------------------------------------------------------------	COMMENT
MEDIUM⚡	finetune/dataset/schema.py	140	# ---------------------------------------------------------------------------	COMMENT
MEDIUM	finetune/jobs/eval_common.py	15	# =============================================================================	COMMENT
MEDIUM	finetune/jobs/eval_common.py	17	# =============================================================================	COMMENT
MEDIUM	finetune/jobs/eval_common.py	242	# =============================================================================	COMMENT
MEDIUM	finetune/jobs/eval_common.py	244	# =============================================================================	COMMENT
MEDIUM	finetune/jobs/eval.py	81	# =============================================================================	COMMENT
MEDIUM	finetune/jobs/eval.py	83	# =============================================================================	COMMENT
MEDIUM	finetune/jobs/eval.py	308	# =============================================================================	COMMENT
MEDIUM	finetune/jobs/eval.py	310	# =============================================================================	COMMENT
MEDIUM	finetune/jobs/eval.py	362	# =============================================================================	COMMENT
MEDIUM	finetune/jobs/eval.py	364	# =============================================================================	COMMENT
MEDIUM	src/index.ts	234	// ── Search ──────────────────────────────────────────────────────────	COMMENT
MEDIUM	src/index.ts	248	// ── Document Retrieval ──────────────────────────────────────────────	COMMENT
MEDIUM	src/index.ts	259	// ── Collection Management ───────────────────────────────────────────	COMMENT
MEDIUM	src/index.ts	276	// ── Context Management ──────────────────────────────────────────────	COMMENT
MEDIUM	src/index.ts	293	// ── Indexing ────────────────────────────────────────────────────────	COMMENT
MEDIUM	src/index.ts	313	// ── Index Health ────────────────────────────────────────────────────	COMMENT
MEDIUM	src/index.ts	321	// ── Lifecycle ───────────────────────────────────────────────────────	COMMENT

Dead Code77 hits · 96 pts

Severity	File	Line	Context
MEDIUM	finetune/train.py	422	CODE
MEDIUM	finetune/train.py	423	CODE
MEDIUM	finetune/train.py	424	CODE
MEDIUM	finetune/train.py	425	CODE
MEDIUM	finetune/train.py	426	CODE
MEDIUM	finetune/train.py	427	CODE
MEDIUM	finetune/train.py	428	CODE
MEDIUM	finetune/train.py	430	CODE
MEDIUM	finetune/train.py	431	CODE
MEDIUM	finetune/train.py	434	CODE
MEDIUM	finetune/train.py	435	CODE
MEDIUM	finetune/train.py	437	CODE
MEDIUM	finetune/train.py	440	CODE
MEDIUM	finetune/train.py	442	CODE
MEDIUM	finetune/train.py	448	CODE
MEDIUM	finetune/train.py	449	CODE
MEDIUM	finetune/train.py	457	CODE
MEDIUM	finetune/train.py	458	CODE
MEDIUM	finetune/train.py	463	CODE
MEDIUM	finetune/train.py	464	CODE
MEDIUM	finetune/train.py	473	CODE
MEDIUM	finetune/train.py	474	CODE
MEDIUM	finetune/train.py	475	CODE
MEDIUM	finetune/train.py	476	CODE
MEDIUM	finetune/train.py	480	CODE
MEDIUM	finetune/train.py	481	CODE
MEDIUM	finetune/train.py	483	CODE
MEDIUM	finetune/train.py	491	CODE
MEDIUM	finetune/train.py	492	CODE
MEDIUM	finetune/train.py	493	CODE
MEDIUM	finetune/train.py	494	CODE
MEDIUM	finetune/train.py	497	CODE
MEDIUM	finetune/train.py	498	CODE
MEDIUM	finetune/train.py	499	CODE
MEDIUM	finetune/train.py	500	CODE
MEDIUM	finetune/train.py	511	CODE
MEDIUM	finetune/train.py	512	CODE
MEDIUM	finetune/train.py	517	CODE
MEDIUM	finetune/train.py	518	CODE
MEDIUM	finetune/train.py	523	CODE
MEDIUM	finetune/train.py	527	CODE
MEDIUM	finetune/train.py	528	CODE
MEDIUM	finetune/train.py	529	CODE
MEDIUM	finetune/train.py	532	CODE
MEDIUM	finetune/train.py	542	CODE
MEDIUM	finetune/train.py	543	CODE
MEDIUM	finetune/train.py	546	CODE
MEDIUM	finetune/train.py	547	CODE
MEDIUM	finetune/train.py	548	CODE
MEDIUM	finetune/train.py	550	CODE
MEDIUM	finetune/train.py	552	CODE
MEDIUM	finetune/train.py	553	CODE
MEDIUM	finetune/train.py	554	CODE
MEDIUM	finetune/train.py	557	CODE
MEDIUM	finetune/train.py	558	CODE
MEDIUM	finetune/train.py	559	CODE
MEDIUM	finetune/train.py	560	CODE
MEDIUM	finetune/train.py	561	CODE
MEDIUM	finetune/train.py	566	CODE
MEDIUM	finetune/train.py	567	CODE
17 more matches not shown…

Hyper-Verbose Identifiers68 hits · 60 pts

Severity	File	Line	Snippet	Context
LOW	tools/oxlint/anti-slop/shared/reflect-method.ts	24	export function isGlobalReflectMethodCall(	CODE
LOW	tools/oxlint/anti-slop/shared/dictionary-types.ts	142	function isEffectivelyEmptyTypeLiteral(type: ESTree.TSTypeLiteral): boolean {	CODE
LOW	tools/oxlint/anti-slop/shared/dictionary-types.ts	146	function isEffectivelyEmptyInterface(	CODE
LOW	tools/oxlint/anti-slop/shared/dictionary-types.ts	158	function resolvedSubstitutionArgument(	CODE
LOW	tools/oxlint/anti-slop/shared/dictionary-types.ts	307	export function classifyUnsafeDictionaryValue(	CODE
LOW	tools/oxlint/anti-slop/shared/dictionary-types.ts	467	export function isPopulatedObjectExpression(expression: ESTree.Expression): boolean {	CODE
LOW	tools/oxlint/anti-slop/shared/dictionary-types.ts	480	export function isKnownEvidenceExpression(expression: ESTree.Expression): boolean {	CODE
LOW	…ols/oxlint/anti-slop/shared/lexical-type-parameters.ts	14	function collectInferTypeParameterNames(	CODE
LOW	…ols/oxlint/anti-slop/shared/lexical-type-parameters.ts	35	export function lexicalTypeParameterNames(	CODE
LOW	tools/oxlint/anti-slop/rules/no-widen-then-assert.ts	18	function unwrapExpressionParentheses(expression: ESTree.Expression): ESTree.Expression {	CODE
LOW	tools/oxlint/anti-slop/rules/no-widen-then-assert.ts	143	function isDefinitelyNarrowerRecordType(type: ESTree.TSType): boolean {	CODE
LOW	tools/oxlint/anti-slop/rules/no-widen-then-assert.ts	171	function resolvedVariableForIdentifier(	CODE
LOW	…s/oxlint/anti-slop/rules/no-chained-type-assertions.ts	6	function isTypeAssertionExpression(node: ESTree.Node): node is TypeAssertionExpression {	CODE
LOW	…s/oxlint/anti-slop/rules/no-chained-type-assertions.ts	10	function unwrapParenthesizedExpression(expression: ESTree.Expression): ESTree.Expression {	CODE
LOW	…s/oxlint/anti-slop/rules/no-chained-type-assertions.ts	27	function isOutermostAssertionInChain(node: TypeAssertionExpression): boolean {	CODE
LOW	…s/oxlint/anti-slop/rules/no-chained-type-assertions.ts	39	function isForbiddenAssertionChain(node: TypeAssertionExpression): boolean {	CODE
LOW	tools/oxlint/anti-slop/rules/no-known-value-widening.ts	125	function isDictionaryAccumulatorTarget(destination: WideningTarget): boolean {	CODE
LOW	…/anti-slop/rules/no-conditional-empty-object-spread.ts	16	function isConditionalEmptyObjectSpread(node: ESTree.Expression): boolean {	CODE
LOW	…ls/oxlint/anti-slop/rules/no-unsafe-dictionary-type.ts	60	function isInsideTypeAliasDeclaration(node: ESTree.Node): boolean {	CODE
LOW	…ols/oxlint/anti-slop/rules/no-shape-in-symbol-names.ts	6	function containsForbiddenSymbolName(name: string): boolean {	CODE
LOW	test/store-cjk-fts.test.ts	165	const startIdx = storeSrc.indexOf("function rebuildFTSForCjkNormalization(");	CODE
LOW	finetune/convert_onnx.py	294	def write_transformers_js_config(onnx_dir: str, quantize_type: str = "q4"):	CODE
LOW⚡	finetune/reward.py	634	def extract_query_from_prompt(prompt: str) -> str:	CODE
LOW	finetune/jobs/eval_common.py	218	def extract_query_from_prompt(prompt):	CODE
LOW	src/trust.ts	106	export function resolveConfigCollectionPath(rawPath: string, configPath: string): string {	CODE
LOW	src/trust.ts	126	export function isCollectionPathInsideProject(configPath: string, rawPath: string): boolean {	CODE
LOW	src/trust.ts	272	export function isLocalConfigTrustOptedIn(env: NodeJS.ProcessEnv = process.env): boolean {	CODE
LOW	src/collections.ts	271	export function getDefaultCollectionNames(): string[] {	CODE
LOW	src/llm.ts	38	export function setNodeLlamaCppModuleForTest(module: NodeLlamaCppModule \| null): void {	CODE
LOW	src/llm.ts	130	export function setLlamaDirWritableForTest(writable: boolean \| undefined): void {	CODE
LOW	src/llm.ts	643	export function resolveParallelismOverride(envValue = process.env.QMD_EMBED_PARALLELISM): number \| undefined {	CODE
LOW	src/llm.ts	686	export function computeGpuContextPoolSize(options: {	CODE
LOW	src/llm.ts	749	async function disposeSequenceThenContext(	CODE
LOW	src/llm.ts	2174	export function isDarwinMetalMitigationActive(): boolean {	CODE
LOW	src/llm.ts	2193	export function isDarwinExitGuardInstalled(): boolean {	CODE
LOW	src/store.ts	357	export function chunkDocumentWithBreakPoints(	CODE
LOW	src/store.ts	503	export function getRelativePathFromPrefix(path: string, prefix: string): string \| null {	CODE
LOW	src/store.ts	832	function createSqliteVecUnavailableError(reason: string): Error {	CODE
LOW	src/store.ts	992	function documentsFtsSchemaIsCurrent(db: Database): boolean {	CODE
LOW	src/store.ts	1085	function rebuildFTSForCjkNormalization(db: Database): void {	CODE
LOW	src/store.ts	1797	function validatePositiveIntegerOption(name: string, value: number \| undefined, fallback: number): number {	CODE
LOW	src/store.ts	1821	function isContentVectorColumnError(error: unknown): boolean {	CODE
LOW	src/store.ts	1840	function ensureContentVectorsStatusIndex(db: Database): void {	CODE
LOW	src/store.ts	1853	function runContentVectorColumnRepairs(db: Database): void {	CODE
LOW	src/store.ts	2520	export function getHashesNeedingEmbedding(db: Database, collection?: string, model: string = DEFAULT_EMBED_MODEL): numbe	CODE
LOW	src/store.ts	2554	export async function maybeAdoptLegacyEmbeddingFingerprint(store: Store, model: string = DEFAULT_EMBED_MODEL): Promise<L	CODE
LOW	src/store.ts	2995	export function findOrMigrateLegacyDocument(	CODE
LOW	src/store.ts	3744	export function getCollectionsWithoutContext(db: Database): { name: string; pwd: string; doc_count: number }[] {	CODE
LOW	src/store.ts	3776	export function getTopLevelPathsWithoutContext(db: Database, collectionName: string): string[] {	CODE
LOW	src/store.ts	4040	function mergeSearchResultsByScore(lists: SearchResult[][], limit: number): SearchResult[] {	CODE
LOW	src/store.ts	4452	function removeIncompleteEmbeddings(db: Database, expectedChunksByHash: Map<string, number>, model: string): number {	CODE
LOW	src/store.ts	4524	export function deleteExpansionCacheEntry(db: Database, query: string, model: string = DEFAULT_QUERY_MODEL): void {	CODE
LOW	src/store.ts	4712	function normalizeLookupPathForIgnore(path: string): string {	CODE
LOW	src/bench/bench.ts	283	export function assertBenchCollectionReady(	CODE
LOW	src/cli/qmd.ts	301	export async function finishSuccessfulCliCommand(options: FinishSuccessfulCliCommandOptions): Promise<void> {	CODE
LOW	src/cli/qmd.ts	494	function sanitizeDiagnosticMessage(message: string): string {	CODE
LOW	src/cli/qmd.ts	749	function localConfigIsFullyTrusted(): boolean {	CODE
LOW	src/cli/qmd.ts	2114	function ensureModelsConfiguredForCli(): { embed: string; generate: string; rerank: string } {	CODE
LOW	src/cli/qmd.ts	2140	export function resolveGenerateModelForCli(): string {	CODE
LOW	src/cli/qmd.ts	3334	function installedSkillStubContent(): string {	CODE
8 more matches not shown…

Verbosity Indicators31 hits · 52 pts

Severity	File	Line	Snippet	Context
LOW⚡	migrate-schema.ts	31	// Step 1: Add collection column to documents	COMMENT
LOW⚡	migrate-schema.ts	36	// Step 2: Populate collection names from collections table	COMMENT
LOW⚡	migrate-schema.ts	47	// Step 3: Verify no NULL values	COMMENT
LOW⚡	migrate-schema.ts	57	// Step 4: Create new documents table without collection_id	COMMENT
LOW	migrate-schema.ts	76	// Step 5: Copy data	COMMENT
LOW⚡	migrate-schema.ts	88	// Step 6: Drop old table and rename new one	COMMENT
LOW⚡	migrate-schema.ts	94	// Step 7: Recreate indices	COMMENT
LOW⚡	migrate-schema.ts	100	// Step 8: Update FTS trigger to use collection name	COMMENT
LOW	finetune/convert_gguf.py	117	# Step 1: Load and merge	COMMENT
LOW⚡	finetune/convert_gguf.py	133	# Step 2: Save merged model	COMMENT
LOW⚡	finetune/convert_gguf.py	139	# Step 3: Setup llama.cpp	COMMENT
LOW⚡	finetune/convert_gguf.py	147	# Step 4: Convert to FP16 GGUF	COMMENT
LOW	finetune/convert_gguf.py	161	# Step 5: Quantize	COMMENT
LOW	finetune/convert_gguf.py	180	# Step 6: Upload	COMMENT
LOW⚡	src/store.ts	5582	// Step 4: RRF fusion — original-query FTS and vector lists get 2x weight;	COMMENT
LOW⚡	src/store.ts	5591	// Step 5: Chunk documents, pick best chunk per doc for reranking.	COMMENT
LOW⚡	src/store.ts	5972	// Step 3: RRF fusion — first list gets 2x weight (assume caller ordered by importance)	COMMENT
LOW⚡	src/store.ts	5982	// Step 4: Chunk documents, pick best chunk per doc for reranking	COMMENT
LOW	src/store.ts	4215	// Step 1: Get vector matches from sqlite-vec (no JOINs allowed).	COMMENT
LOW	src/store.ts	4249	// Step 2: Get chunk info and document data	COMMENT
LOW	src/store.ts	5471	// Step 1: BM25 probe — strong signal skips expensive LLM expansion	COMMENT
LOW	src/store.ts	5485	// Step 2: Expand query (or skip if strong signal)	COMMENT
LOW	src/store.ts	5504	// Step 3: Route searches by query type	COMMENT
LOW	src/store.ts	5668	// Step 6: Rerank chunks (NOT full bodies)	COMMENT
LOW	src/store.ts	5682	// Step 7: Blend RRF position score with reranker score	COMMENT
LOW	src/store.ts	5734	// Step 8: Dedup by file (safety net — prevents duplicate output)	COMMENT
LOW	src/store.ts	5908	// Step 1: Run FTS for all lex searches (sync, instant)	COMMENT
LOW	src/store.ts	5929	// Step 2: Batch embed and run vector searches for vec/hyde	COMMENT
LOW	src/store.ts	6062	// Step 5: Rerank chunks	COMMENT
LOW	src/store.ts	6076	// Step 6: Blend RRF position score with reranker score	COMMENT
LOW	src/store.ts	6127	// Step 7: Dedup by file	COMMENT

Structural Annotation Overuse31 hits · 52 pts

Severity	File	Line	Snippet	Context
LOW⚡	migrate-schema.ts	31	// Step 1: Add collection column to documents	COMMENT
LOW⚡	migrate-schema.ts	36	// Step 2: Populate collection names from collections table	COMMENT
LOW⚡	migrate-schema.ts	47	// Step 3: Verify no NULL values	COMMENT
LOW⚡	migrate-schema.ts	57	// Step 4: Create new documents table without collection_id	COMMENT
LOW	migrate-schema.ts	76	// Step 5: Copy data	COMMENT
LOW⚡	migrate-schema.ts	88	// Step 6: Drop old table and rename new one	COMMENT
LOW⚡	migrate-schema.ts	94	// Step 7: Recreate indices	COMMENT
LOW⚡	migrate-schema.ts	100	// Step 8: Update FTS trigger to use collection name	COMMENT
LOW	finetune/convert_gguf.py	117	# Step 1: Load and merge	COMMENT
LOW⚡	finetune/convert_gguf.py	133	# Step 2: Save merged model	COMMENT
LOW⚡	finetune/convert_gguf.py	139	# Step 3: Setup llama.cpp	COMMENT
LOW⚡	finetune/convert_gguf.py	147	# Step 4: Convert to FP16 GGUF	COMMENT
LOW	finetune/convert_gguf.py	161	# Step 5: Quantize	COMMENT
LOW	finetune/convert_gguf.py	180	# Step 6: Upload	COMMENT
LOW⚡	src/store.ts	5582	// Step 4: RRF fusion — original-query FTS and vector lists get 2x weight;	COMMENT
LOW⚡	src/store.ts	5591	// Step 5: Chunk documents, pick best chunk per doc for reranking.	COMMENT
LOW⚡	src/store.ts	5972	// Step 3: RRF fusion — first list gets 2x weight (assume caller ordered by importance)	COMMENT
LOW⚡	src/store.ts	5982	// Step 4: Chunk documents, pick best chunk per doc for reranking	COMMENT
LOW	src/store.ts	4215	// Step 1: Get vector matches from sqlite-vec (no JOINs allowed).	COMMENT
LOW	src/store.ts	4249	// Step 2: Get chunk info and document data	COMMENT
LOW	src/store.ts	5471	// Step 1: BM25 probe — strong signal skips expensive LLM expansion	COMMENT
LOW	src/store.ts	5485	// Step 2: Expand query (or skip if strong signal)	COMMENT
LOW	src/store.ts	5504	// Step 3: Route searches by query type	COMMENT
LOW	src/store.ts	5668	// Step 6: Rerank chunks (NOT full bodies)	COMMENT
LOW	src/store.ts	5682	// Step 7: Blend RRF position score with reranker score	COMMENT
LOW	src/store.ts	5734	// Step 8: Dedup by file (safety net — prevents duplicate output)	COMMENT
LOW	src/store.ts	5908	// Step 1: Run FTS for all lex searches (sync, instant)	COMMENT
LOW	src/store.ts	5929	// Step 2: Batch embed and run vector searches for vec/hyde	COMMENT
LOW	src/store.ts	6062	// Step 5: Rerank chunks	COMMENT
LOW	src/store.ts	6076	// Step 6: Blend RRF position score with reranker score	COMMENT
LOW	src/store.ts	6127	// Step 7: Dedup by file	COMMENT

Excessive Try-Catch Wrapping19 hits · 22 pts

Severity	File	Line	Snippet	Context
LOW	finetune/train_unsloth.py	169	except Exception as e:	CODE
LOW	finetune/train_unsloth.py	180	except Exception as e:	CODE
MEDIUM	finetune/eval_retrieval.py	368	print(f"Error: Golden data file not found: {args.golden}", file=sys.stderr)	CODE
LOW	finetune/train.py	279	except Exception:	CODE
LOW	finetune/train.py	409	except Exception:	CODE
LOW	finetune/train.py	454	except Exception:	CODE
LOW	finetune/train.py	628	except Exception:	CODE
LOW	finetune/train.py	644	except Exception:	CODE
LOW	finetune/dataset/validate_schema.py	38	except Exception as e:	CODE
MEDIUM	finetune/dataset/prepare_data.py	111	print(f"Error: No files found matching: {args.input}")	CODE
MEDIUM	finetune/dataset/prepare_data.py	117	print(f"Error: Input file not found: {input_path}")	CODE
MEDIUM	finetune/dataset/analyze_data.py	218	print(f"Error: Could not find dataset at {input_path}")	CODE
LOW⚡	finetune/dataset/schema.py	133	except Exception as e:	CODE
LOW	finetune/dataset/schema.py	186	except Exception:	CODE
LOW	finetune/dataset/schema.py	220	except Exception:	CODE
LOW	finetune/experiments/gepa/dspy_gepa.py	100	except Exception:	CODE
MEDIUM	finetune/experiments/gepa/dspy_gepa.py	140	print("Error: DSPy expects provider/model format for LM strings (e.g., xai/grok-4-1-fast-reasoning).")	CODE
LOW	finetune/experiments/gepa/example.py	114	except Exception as exc:	CODE
LOW	finetune/experiments/gepa/score.py	32	except Exception:	CODE

Deep Nesting18 hits · 18 pts

Severity	File	Line	Context
LOW	finetune/train_unsloth.py	26	CODE
LOW	finetune/benchmark.py	62	CODE
LOW	finetune/eval_retrieval.py	79	CODE
LOW	finetune/eval_retrieval.py	115	CODE
LOW	finetune/eval_retrieval.py	332	CODE
LOW	finetune/reward.py	90	CODE
LOW	finetune/reward.py	140	CODE
LOW	finetune/reward.py	267	CODE
LOW	finetune/reward.py	408	CODE
LOW	finetune/dataset/analyze_data.py	84	CODE
LOW	finetune/dataset/analyze_data.py	196	CODE
LOW	finetune/dataset/schema.py	142	CODE
LOW	finetune/experiments/gepa/generate.py	32	CODE
LOW	finetune/jobs/eval_common.py	42	CODE
LOW	finetune/jobs/eval_common.py	67	CODE
LOW	finetune/jobs/eval.py	108	CODE
LOW	finetune/jobs/eval.py	133	CODE
LOW	finetune/jobs/eval.py	204	CODE

Unused Imports17 hits · 16 pts

Severity	File	Line	Context
LOW	finetune/train_unsloth.py	14	CODE
LOW	finetune/train_unsloth.py	67	CODE
LOW	finetune/train_unsloth.py	111	CODE
LOW	finetune/train.py	44	CODE
LOW	finetune/convert_onnx.py	44	CODE
LOW	finetune/convert_onnx.py	45	CODE
LOW	finetune/dataset/validate_schema.py	8	CODE
LOW	finetune/dataset/score_data.py	8	CODE
LOW	finetune/dataset/prepare_data_lfm2.py	14	CODE
LOW	finetune/dataset/schema.py	16	CODE
LOW	finetune/experiments/gepa/generate.py	4	CODE
LOW	finetune/experiments/gepa/dspy_gepa.py	4	CODE
LOW	finetune/experiments/gepa/example.py	4	CODE
LOW	finetune/experiments/gepa/optimizer.py	4	CODE
LOW	finetune/experiments/lfm2/sft_lfm2.py	38	CODE
LOW	finetune/jobs/eval.py	23	CODE
LOW	finetune/jobs/eval.py	26	CODE

Over-Commented Block17 hits · 16 pts

Severity	File	Line	Snippet	Context
LOW	example-index.yml	1	# QMD Collections Configuration	COMMENT
LOW	test/smoke-install.sh	1	#!/usr/bin/env bash	COMMENT
LOW	test/llm.test.ts	941	// Performance is machine/load dependent. We only assert batch isn't drastically worse.	COMMENT
LOW	test/store-cjk-fts.test.ts	181	// API. Any .all() used for the source-body scan must be paired with a	COMMENT
LOW	finetune/convert_gguf.py	1	#!/usr/bin/env python3	COMMENT
LOW	finetune/train.py	1	# /// script	COMMENT
LOW	finetune/convert_onnx.py	1	#!/usr/bin/env python3	COMMENT
LOW	finetune/experiments/lfm2/sft_lfm2.yaml	1	# SFT Training Config for QMD Query Expansion with LiquidAI LFM2	COMMENT
LOW	finetune/experiments/lfm2/sft_lfm2.py	1	# /// script	COMMENT
LOW	finetune/experiments/grpo/grpo.py	1	# /// script	COMMENT
LOW	finetune/jobs/sft.py	1	# /// script	COMMENT
LOW	scripts/extract-changelog.sh	1	#!/usr/bin/env bash	COMMENT
LOW	scripts/release.sh	1	#!/usr/bin/env bash	COMMENT
LOW	src/llm.ts	2141	// libggml-metal on macOS keeps allocated model memory wired via "residency	COMMENT
LOW	src/store.ts	2761	return 0;	COMMENT
LOW	src/store.ts	3001	const existing = findActiveDocument(db, collectionName, path);	COMMENT
LOW	src/mcp/server.ts	801	// Opt into production mode when the MCP server is actually started, not	COMMENT

Fake / Example Data9 hits · 9 pts

Severity	File	Line	Snippet	Context
LOW	test/store.test.ts	2667	const padLine = "Lorem ipsum dolor sit amet\n";	CODE
LOW	test/store.test.ts	2667	const padLine = "Lorem ipsum dolor sit amet\n";	CODE
LOW	test/store.test.ts	2681	const padding = "Lorem ipsum dolor sit amet\n".repeat(200);	CODE
LOW	test/store.test.ts	2681	const padding = "Lorem ipsum dolor sit amet\n".repeat(200);	CODE
LOW	test/mcp.test.ts	165	body: "# Large Document\n\n" + "Lorem ipsum ".repeat(2000), // ~24KB	CODE
LOW	test/ast-chunking.test.ts	116	sections.push(`# Section ${i}\n\n${"Lorem ipsum dolor sit amet. ".repeat(40)}\n`);	CODE
LOW	test/ast-chunking.test.ts	116	sections.push(`# Section ${i}\n\n${"Lorem ipsum dolor sit amet. ".repeat(40)}\n`);	CODE
LOW	test/store-cjk-fts.test.ts	229	const bigBody = "lorem ipsum dolor sit amet ".repeat(450); // ~12 KB	CODE
LOW	test/store-cjk-fts.test.ts	229	const bigBody = "lorem ipsum dolor sit amet ".repeat(450); // ~12 KB	CODE

Modern AI Meta-Vocabulary2 hits · 6 pts

Severity	File	Line	Snippet	Context
MEDIUM	finetune/experiments/lfm2/sft_lfm2.yaml	5	# Recommended for: agentic tasks, data extraction, RAG, creative writing.	COMMENT
MEDIUM	finetune/experiments/lfm2/sft_lfm2.yaml	5	# Recommended for: agentic tasks, data extraction, RAG, creative writing.	COMMENT

Cross-Language Confusion1 hit · 5 pts

Severity	File	Line	Snippet	Context
HIGH	finetune/train.py	419	print(" cd finetune && uv run python experiments/grpo/grpo.py")	CODE

Self-Referential Comments1 hit · 4 pts

Severity	File	Line	Snippet	Context
MEDIUM⚡	example-index.yml	4	# This file is the single source of truth for your collections. Every	COMMENT

Example Usage Blocks1 hit · 2 pts

Severity	File	Line	Snippet	Context
LOW	test/smoke-install.sh	6	# Usage:	COMMENT

Redundant / Tautological Comments1 hit · 1 pts

Severity	File	Line	Snippet	Context
LOW	finetune/train.py	266	# Check if output looks like a HF Hub path (contains /)	COMMENT

Analysis Overview

What These Metrics Mean

Score History

Severity Breakdown

Directory Score Breakdown

Pattern Findings