1.2

Adjusted Signal

1.2

Raw Pattern Score

100%

Time Factor

2026-10-31

Last Push

7.3K

Stars

C++

Language

796.3K

Lines of Code

2.0K

Files

972

Pattern Hits

2026-10-31

Scan Date

0.00

HC Hit Rate

Risk Review · Payload v1

Use each vector to guide manual verification of the underlying findings.

Payload schema: 1.0.0 · Engine: 2.0.0 · Scoring: heuristic-density-v1 · Scan date: 2026-10-31.

Cognitive Ownership Risk 5.6/100 · LOW

Can maintainers explain intent, assumptions, and trade-offs? 895 finding(s).

Hallucinated Dependency & Supply Chain Risk 0.5/100 · LOW

Are imports declared and package/API names verified? 76 finding(s).

Agent Over-Engineering Ratio 0.0/100 · LOW

Are wrappers, guards, and repeated structures necessary? 2 finding(s).

PR Review Velocity Impact 6.0/100 · LOW

Which changes need focused human review? 972 finding(s).

What These Metrics Mean

Adjusted Signal: Legacy pattern score normalized per 1,000 scanned lines and multiplied by the repository-level time factor. It is not an estimate of AI authorship.
Raw Pattern Score: Sum of severity- and context-weighted pattern findings before applying the time factor.
Time Factor: A coarse multiplier based on repository last-push metadata. It does not identify when individual code was written.
Pattern Hits: Count of configured pattern findings. Review each location and surrounding code before acting.
HC Hit Rate: High- and critical-severity findings per scanned file. Severity is a configured detector weight, not proof of a vulnerability or authorship.
Lines / Files: Scan coverage denominators used by the legacy normalized score.

Pattern-Signal History

This chart tracks the legacy adjusted pattern score across scan runs. Changes can reflect source changes, scan coverage, or detector/scoring versions; they do not measure AI adoption or prove that a team refactored code. Compare runs only when their scoring versions are compatible. The dashed line tracks the raw pattern-hit count.

Severity Breakdown

Groups findings by configured severity weight. Severity is not a confidence estimate, security rating, or proof that a package is missing. Verify imports against the project's manifests and inspect each finding in its local design context.

CRITICAL 0HIGH 0MEDIUM 8LOW 964

Directory Score Breakdown

This horizontal bar chart shows where configured pattern scores accumulate across directories. Treat the bars as locations to inspect, not as evidence of authorship or code quality. Use the file-level findings to decide whether a directory needs a focused ownership, dependency, complexity, or review check.

Pattern Findings

The scanner identified 972 distinct pattern matches across 10 syntactic categories. Each entry below represents a discrete location in the source code where the engine recorded a detector match to inspect in context. 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 configured finding weight (CRITICAL / HIGH / MEDIUM / LOW). The Context column identifies whether the match occurred inside executable code, an inline comment, or a string literal — legacy comment-context matches receive a ×1.5 score weight. The ⚡ bolt icon marks clustered matches: three or more patterns within a 10-line window, each receiving an additional ×1.5 density multiplier under the legacy scoring model.

Over-Commented Block864 hits · 860 pts

Severity	File	Line	Snippet	Context
LOW	test/case.h	1	#ifndef _GUARD_case_h_	COMMENT
LOW	test/case.h	41	* ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER	COMMENT
LOW	test/unit/nnc/dynamic.graph.tests.c	801	// 10, 11	COMMENT
LOW	test/int/nnc/cudnn.tests.c	1	#include "case.h"	COMMENT
LOW	test/int/nnc/moe_weights_streaming.tests.c	1	#include "case.h"	COMMENT
LOW	test/int/nnc/mps.thread.tests.c	1	#include "case.h"	COMMENT
LOW	test/int/nnc/mpsblas.tests.c	1	#include "case.h"	COMMENT
LOW	test/int/nnc/mpsdnn.tests.c	1	#include "case.h"	COMMENT
LOW	test/int/nnc/tensor.tests.c	1	#include "ccv.h"	COMMENT
LOW	test/int/nnc/packed_expert_addressing.tests.c	1	#include "case.h"	COMMENT
LOW	test/int/nnc/scaled_dot_product_arg_partition.tests.c	1	#include "case.h"	COMMENT
LOW	bin/dpmvldtr.rb	1	#!/usr/bin/env ruby	COMMENT
LOW	bin/tld.c	1	#include <ccv.h>	COMMENT
LOW	bin/nnc/square_attention_test.cpp	1	extern "C" {	COMMENT
LOW	bin/nnc/ane_quant_fusion_probe.cpp	1	// Compare the original two-dispatch ANE preparation with fused candidates.	COMMENT
LOW	bin/nnc/gemm_kernel_gen.cpp	1	extern "C" {	COMMENT
LOW	bin/nnc/laplacian_test.cpp	1	extern "C" {	COMMENT
LOW	bin/nnc/laplacian_test.cpp	141	NS::SharedPtr<MTL::Buffer> bufferC = NS::TransferPtr(device->newBuffer(C, descriptor.memoryPrecisions.C.size() * probl	COMMENT
LOW	bin/nnc/laplacian_test.cpp	161	// problemSize = 1489 \| A B \| 768 threads/core \| 7902 GFLOPS	COMMENT
LOW	bin/nnc/laplacian_test.cpp	181	// problemSize = 1488 \| A^T B^T \| 1024 threads/core \| 9024 GFLOPS	COMMENT
LOW	bin/nnc/laplacian_test.cpp	201	// problemSize = 1488 \| A^T B^T \| 1024 threads/core \| 3289 GFLOPS	COMMENT
LOW	bin/nnc/attention_kernel_gen.cpp	1	extern "C" {	COMMENT
LOW	bin/nnc/attention_kernel_gen.cpp	81	// L \| 2e-5 \| 7e-3 (FP16) \|	COMMENT
LOW	bin/nnc/attention_kernel_gen.cpp	101	// \| Maximum \| +0.2% \| +4.4% \| +5.6% \|	COMMENT
LOW	bin/nnc/attention_kernel_gen.cpp	121	// \| Minimum \| -1.9% \| -1.5% \| -1.5% \|	COMMENT
LOW	bin/nnc/attention_kernel_gen.cpp	141	//	COMMENT
LOW	bin/nnc/attention_kernel_gen.cpp	161	// was without utilizing the native hardware instructions for BF16 to	COMMENT
LOW	bin/nnc/attention_kernel_gen.cpp	241	// throughput. S = Q * K is the only place where it can be employed	COMMENT
LOW	bin/nnc/rowwise_int8_gemm_bench.c	1	#include <ccv.h>	COMMENT
LOW	bin/nnc/ane_hadamard_bench.c	1	#include <ccv.h>	COMMENT
LOW	bin/nnc/opt/conv-opt.c	1	#include <ccv.h>	COMMENT
LOW	bin/nnc/opt/auto-tune-conv.c	1	#include <ccv.h>	COMMENT
LOW	bin/nnc/opt/conv-gemm.c	1	#include <ccv.h>	COMMENT
LOW	bin/nnc/cudnn/conv-cudnn-forw.c	1	#include <ccv.h>	COMMENT
LOW	bin/nnc/cudnn/conv-cudnn-back.c	1	#include <ccv.h>	COMMENT
LOW	bin/mfa/hyper_connection_bench.cpp	1	#include <algorithm>	COMMENT
LOW	bin/mfa/walsh_hadamard_transform_bench.mm	1	#include <algorithm>	COMMENT
LOW	bin/mfa/gated_delta_bench.cpp	1	#include <algorithm>	COMMENT
LOW	bin/mfa/sdpa_backward_compare_bench.mm	1	#include <algorithm>	COMMENT
LOW	bin/mfa/na_int8_attention_float_pv_bench.cpp	1	// Three-way forward prototype: INT8 PV, floating PV with INT8 QK, NAAttention.	COMMENT
LOW	bin/mfa/na_attention_decode1_bench.cpp	1	#include <algorithm>	COMMENT
LOW	bin/mfa/na_int8_attention_kcenter_replay.cpp	1	// Replay original FP16 Q/K snapshots through the production centered quantizers.	COMMENT
LOW	bin/mfa/na_int8_v_mean_bench.cpp	1	// Compare the production V-mean pipeline with the previous single-pass reduction.	COMMENT
LOW	bin/mfa/na_attention_backward_bench.cpp	1	#include <algorithm>	COMMENT
LOW	bin/mfa/conv3d_layout_scaffold.cpp	1	#include <algorithm>	COMMENT
LOW	bin/mfa/moe_routing_bench.cpp	1	#include <algorithm>	COMMENT
LOW	bin/mfa/na_int8_rowwise_x_bench.cpp	1	#include <algorithm>	COMMENT
LOW	bin/mfa/na_int8_rowwise_x_bench.cpp	21	}	COMMENT
LOW	bin/mfa/na_sparse_indexed_attention_bench.cpp	1	#include <algorithm>	COMMENT
LOW	bin/mfa/transpose_bench.cpp	1	#include <algorithm>	COMMENT
LOW	bin/mfa/int8_matmul_tile_bench.cpp	1	// Tile sweep for the dense Int8MatMul without neural accelerators (Apple9:	COMMENT
LOW	bin/mfa/int8_matmul_tile_bench.cpp	21	#include <cstdint>	COMMENT
LOW	bin/mfa/int8_matmul_bench.cpp	1	#include <algorithm>	COMMENT
LOW	bin/mfa/index_select_bench.cpp	1	#include <algorithm>	COMMENT
LOW	bin/mfa/attention_backward_compare_probe.cpp	1	#include <algorithm>	COMMENT
LOW	bin/mfa/dump_na_attention_source.cpp	1	#include <cmath>	COMMENT
LOW	bin/mfa/dynamic_m_bench.cpp	1	// Build: make dynamic_m_bench. Reports median GPU time including activation quantization.	COMMENT
LOW	bin/mfa/na_int8_attention_forward_bench.cpp	1	// GPU timings through the production INT8 attention dispatcher, including Q/K/V quantization.	COMMENT
LOW	bin/mfa/na_int8_matmul_forward_bench.cpp	1	// GPU timings through the shipping INT8 GEMM dispatcher, including activation quantization.	COMMENT
LOW	bin/mfa/attention_r1_bench.cpp	1	// Production decode timing, randomized CPU validation, and optional stage profiling.	COMMENT
804 more matches not shown…

Unused Imports76 hits · 76 pts

Severity	File	Line	Context
LOW	test/unit/.ycm_extra_conf.py	2	CODE
LOW	test/unit/.ycm_extra_conf.py	3	CODE
LOW	test/int/.ycm_extra_conf.py	2	CODE
LOW	test/int/.ycm_extra_conf.py	3	CODE
LOW	bin/.ycm_extra_conf.py	2	CODE
LOW	bin/.ycm_extra_conf.py	3	CODE
LOW	bin/nnc/.ycm_extra_conf.py	2	CODE
LOW	bin/nnc/.ycm_extra_conf.py	3	CODE
LOW	bin/nnc/cudnn/.ycm_extra_conf.py	2	CODE
LOW	bin/nnc/cudnn/.ycm_extra_conf.py	3	CODE
LOW	lib/.ycm_extra_conf.py	1	CODE
LOW	lib/.ycm_extra_conf.py	2	CODE
LOW	lib/.ycm_extra_conf.py	3	CODE
LOW	lib/nnc/.ycm_extra_conf.py	2	CODE
LOW	lib/nnc/.ycm_extra_conf.py	3	CODE
LOW	lib/nnc/cmd/gelu/gpu/.ycm_extra_conf.py	2	CODE
LOW	lib/nnc/cmd/gelu/gpu/.ycm_extra_conf.py	3	CODE
LOW	lib/nnc/cmd/upsample/mps/.ycm_extra_conf.py	2	CODE
LOW	lib/nnc/cmd/upsample/mps/.ycm_extra_conf.py	3	CODE
LOW	lib/nnc/cmd/blas/gpu/.ycm_extra_conf.py	2	CODE
LOW	lib/nnc/cmd/blas/gpu/.ycm_extra_conf.py	3	CODE
LOW	lib/nnc/cmd/blas/mps/.ycm_extra_conf.py	2	CODE
LOW	lib/nnc/cmd/blas/mps/.ycm_extra_conf.py	3	CODE
LOW	lib/nnc/cmd/util/gpu/.ycm_extra_conf.py	2	CODE
LOW	lib/nnc/cmd/util/gpu/.ycm_extra_conf.py	3	CODE
LOW	lib/nnc/cmd/loss/gpu/.ycm_extra_conf.py	2	CODE
LOW	lib/nnc/cmd/loss/gpu/.ycm_extra_conf.py	3	CODE
LOW	…md/scaled_dot_product_attention/gpu/.ycm_extra_conf.py	2	CODE
LOW	…md/scaled_dot_product_attention/gpu/.ycm_extra_conf.py	3	CODE
LOW	lib/nnc/cmd/ew/gpu/.ycm_extra_conf.py	2	CODE
LOW	lib/nnc/cmd/ew/gpu/.ycm_extra_conf.py	3	CODE
LOW	lib/nnc/cmd/ew/mps/.ycm_extra_conf.py	2	CODE
LOW	lib/nnc/cmd/ew/mps/.ycm_extra_conf.py	3	CODE
LOW	lib/nnc/cmd/softmax_loss/gpu/.ycm_extra_conf.py	2	CODE
LOW	lib/nnc/cmd/softmax_loss/gpu/.ycm_extra_conf.py	3	CODE
LOW	lib/nnc/cmd/comm/gpu/.ycm_extra_conf.py	2	CODE
LOW	lib/nnc/cmd/comm/gpu/.ycm_extra_conf.py	3	CODE
LOW	lib/nnc/cmd/norm/gpu/.ycm_extra_conf.py	2	CODE
LOW	lib/nnc/cmd/norm/gpu/.ycm_extra_conf.py	3	CODE
LOW	lib/nnc/cmd/tanh/gpu/.ycm_extra_conf.py	2	CODE
LOW	lib/nnc/cmd/tanh/gpu/.ycm_extra_conf.py	3	CODE
LOW	lib/nnc/cmd/sgd/gpu/.ycm_extra_conf.py	2	CODE
LOW	lib/nnc/cmd/sgd/gpu/.ycm_extra_conf.py	3	CODE
LOW	lib/nnc/cmd/dropout/gpu/.ycm_extra_conf.py	2	CODE
LOW	lib/nnc/cmd/dropout/gpu/.ycm_extra_conf.py	3	CODE
LOW	lib/nnc/cmd/compression/gpu/.ycm_extra_conf.py	2	CODE
LOW	lib/nnc/cmd/compression/gpu/.ycm_extra_conf.py	3	CODE
LOW	lib/nnc/cmd/softmax/gpu/.ycm_extra_conf.py	2	CODE
LOW	lib/nnc/cmd/softmax/gpu/.ycm_extra_conf.py	3	CODE
LOW	lib/nnc/cmd/sigmoid/gpu/.ycm_extra_conf.py	2	CODE
LOW	lib/nnc/cmd/sigmoid/gpu/.ycm_extra_conf.py	3	CODE
LOW	lib/nnc/cmd/index/gpu/.ycm_extra_conf.py	2	CODE
LOW	lib/nnc/cmd/index/gpu/.ycm_extra_conf.py	3	CODE
LOW	lib/nnc/cmd/index/mps/.ycm_extra_conf.py	2	CODE
LOW	lib/nnc/cmd/index/mps/.ycm_extra_conf.py	3	CODE
LOW	lib/nnc/cmd/sigmoid_loss/gpu/.ycm_extra_conf.py	2	CODE
LOW	lib/nnc/cmd/sigmoid_loss/gpu/.ycm_extra_conf.py	3	CODE
LOW	lib/nnc/cmd/rand/gpu/.ycm_extra_conf.py	2	CODE
LOW	lib/nnc/cmd/rand/gpu/.ycm_extra_conf.py	3	CODE
LOW	lib/nnc/cmd/convolution/gpu/.ycm_extra_conf.py	2	CODE
16 more matches not shown…

AI Slop Vocabulary6 hits · 17 pts

Severity	File	Line	Snippet	Context
MEDIUM	lib/ccv.h	2162	// approach can generate more robust feature due to the overfitting	COMMENT
MEDIUM	lib/nnc/ccv_nnc_symbolic_graph_compile.c	211	// to fully utilize memory.	COMMENT
MEDIUM	lib/nnc/ccv_nnc_micro_simplify.c	233	// Essentially, we don't need to move statements, loop-carried variables, just the loop id and the start / end index.	COMMENT
MEDIUM	…ss/include/cutlass/gemm/kernel/tile_scheduler_params.h	773	// the problem can leverage a basic split-K decomposition for the stream-K tiles.	COMMENT
MEDIUM	lib/nnc/mfa/kernels/GEMMDescriptor.cpp	190	// Definitely NOT a global flag that forces all matrices to change from	COMMENT
MEDIUM	site/0.6/doc/doc-convnet/index.html	111	<p>See <a href="http://www.image-net.org/challenges/LSVRC/2013/results.php#cls">http://www.image-net.org/challenges/LSVR	CODE

Verbosity Indicators10 hits · 16 pts

Severity	File	Line	Snippet	Context
LOW	lib/nnc/ccv_cnnp_model_addons.c	499	// Otherwise, we need to check if it is permute. For permute, we cannot do alias directly.	COMMENT
LOW	lib/nnc/ccv_cnnp_model_addons.c	634	// Otherwise, we need to check if it is permute. For permute, we cannot do alias directly.	COMMENT
LOW	lib/nnc/ccv_cnnp_model_addons.c	644	// If the new dim has different nd, or we actually have a stride, we need to check if it is no permute or not.	COMMENT
LOW	…/cutlass/epilogue/threadblock/fusion/visitor_store.hpp	561	// Step 1: reduce fragment input (Src) into tRS_rSrc	COMMENT
LOW	…/cutlass/epilogue/threadblock/fusion/visitor_store.hpp	572	// Step 2: copy the partial results in tRS_rSrc to sRows in shared memory	COMMENT
LOW	…/cutlass/epilogue/threadblock/fusion/visitor_store.hpp	584	// Step 3: copy the partial results in sRows to tSR_sRow for reduction	COMMENT
LOW	…/cutlass/epilogue/threadblock/fusion/visitor_store.hpp	598	// Step 4: atomically reduce the results to global memory	COMMENT
LOW	…/cutlass/transform/collective/sm90_wgmma_transpose.hpp	341	/// Step 0: Step 1:	COMMENT
LOW⚡	…/cutlass/transform/collective/sm90_wgmma_transpose.hpp	559	/// Step 0: Step 1: Step 2: Step 3:	COMMENT
LOW⚡	…/cutlass/transform/collective/sm90_wgmma_transpose.hpp	569	/// Step 4: Step 5: Step 6: Step 7:	COMMENT

Structural Annotation Overuse7 hits · 12 pts

Severity	File	Line	Snippet	Context
LOW	…/cutlass/epilogue/threadblock/fusion/visitor_store.hpp	561	// Step 1: reduce fragment input (Src) into tRS_rSrc	COMMENT
LOW	…/cutlass/epilogue/threadblock/fusion/visitor_store.hpp	572	// Step 2: copy the partial results in tRS_rSrc to sRows in shared memory	COMMENT
LOW	…/cutlass/epilogue/threadblock/fusion/visitor_store.hpp	584	// Step 3: copy the partial results in sRows to tSR_sRow for reduction	COMMENT
LOW	…/cutlass/epilogue/threadblock/fusion/visitor_store.hpp	598	// Step 4: atomically reduce the results to global memory	COMMENT
LOW	…/cutlass/transform/collective/sm90_wgmma_transpose.hpp	341	/// Step 0: Step 1:	COMMENT
LOW⚡	…/cutlass/transform/collective/sm90_wgmma_transpose.hpp	559	/// Step 0: Step 1: Step 2: Step 3:	COMMENT
LOW⚡	…/cutlass/transform/collective/sm90_wgmma_transpose.hpp	569	/// Step 4: Step 5: Step 6: Step 7:	COMMENT

Deep Nesting5 hits · 5 pts

Severity	File	Line	Context
LOW	bin/mfa/analyze_krea_qk.py	28	CODE
LOW	bin/mfa/analyze_krea_qk.py	134	CODE
LOW	bin/mfa/analyze_krea_qk.py	219	CODE
LOW	bin/mfa/attention_r1_sweep.py	27	CODE
LOW	…marks/na-int8-quantize-pv-bank-20261010/launch_ipad.py	13	CODE

Self-Referential Comments1 hit · 3 pts

Severity	File	Line	Snippet	Context
MEDIUM	lib/nnc/gpu/3rdparty/flash_attn/src/generate_kernels.py	3	# This file is run to generate the kernel instantiations for the flash_attn kernels	COMMENT

Excessive Try-Catch Wrapping1 hit · 2 pts

Severity	File	Line	Snippet	Context
MEDIUM	lib/nnc/mfa/packager.py	50	print("Error: File not found or could not be read.")	CODE

Example Usage Blocks1 hit · 2 pts

Severity	File	Line	Snippet	Context
LOW	site/_plugins/debug.rb	2	# Usage:	COMMENT

AI Structural Patterns1 hit · 1 pts

Severity	File	Line	Snippet	Context
LOW	…s/na-int8-quantize-pv-bank-20261010/check_crossover.py	31		CODE

liuliu/ccv

Open-Source Code Quality, Security and Technical Debt Review