2.3

Adjusted Signal

2.3

Raw Pattern Score

100%

Time Factor

2026-10-30

Last Push

3.4K

Stars

C++

Language

682.6K

Lines of Code

2.0K

Files

1.2K

Pattern Hits

2026-10-31

Scan Date

0.02

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 8.7/100 · LOW

Can maintainers explain intent, assumptions, and trade-offs? 1096 finding(s).

Hallucinated Dependency & Supply Chain Risk 0.5/100 · LOW

Are imports declared and package/API names verified? 55 finding(s).

Agent Over-Engineering Ratio 3.5/100 · LOW

Are wrappers, guards, and repeated structures necessary? 323 finding(s).

PR Review Velocity Impact 10.8/100 · LOW

Which changes need focused human review? 1235 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 41MEDIUM 59LOW 1135

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 1235 distinct pattern matches across 24 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 Block575 hits · 569 pts

Severity	File	Line	Snippet	Context
LOW	convert_hf_to_gguf.py	1821	assert self.block_count == len(_block_configs)	COMMENT
LOW	convert_hf_to_gguf_update.py	1	#!/usr/bin/env python3	COMMENT
LOW	convert_hf_to_gguf_update.py	321	CHK_TXT,	COMMENT
LOW	ci/run.sh	1	#/bin/bash	COMMENT
LOW	include/llama.h	1	//	COMMENT
LOW	include/llama.h	21	# ifdef LLAMA_BUILD	COMMENT
LOW	include/llama.h	41	#define LLAMA_DEFAULT_SEED 0xFFFFFFFF	COMMENT
LOW	include/llama.h	81	LLAMA_VOCAB_TYPE_RWKV = 5, // RWKV tokenizer based on greedy tokenization	COMMENT
LOW	include/llama.h	101	// LLAMA_VOCAB_PRE_TYPE_SMAUG = 14,	COMMENT
LOW	include/llama.h	1021	struct llama_context * ctx,	COMMENT
LOW	include/llama.h	1141	// Return batch for single sequence of tokens starting at pos_0	COMMENT
LOW	include/llama.h	1201	// Set whether to use causal attention or not	COMMENT
LOW	include/llama.h	1241		COMMENT
LOW	include/llama.h	1341	llama_token token,	COMMENT
LOW	gguf-py/scripts/gguf_set_metadata.py	21	return	COMMENT
LOW	tests/test-jinja.cpp	1	#include <string>	COMMENT
LOW	tests/test-grammar-llguidance.cpp	541	},	COMMENT
LOW	tests/test-rope.cpp	1	#include "ggml.h"	COMMENT
LOW	tests/test-rope.cpp	21	#define MIN(a, b) ((a) < (b) ? (a) : (b))	COMMENT
LOW	tests/test-chat.cpp	1	// Tests chat handling, including grammar generation and parsing for tool calling, for various templates.	COMMENT
LOW	tests/CMakeLists.txt	161	# target_include_directories(test-json-schema-to-grammar PRIVATE ${CMAKE_CURRENT_SOURCE_DIR}/../examples/server)	COMMENT
LOW	tests/test-grad0.cpp	1	#define _CRT_SECURE_NO_DEPRECATE // Disables ridiculous "unsafe" warnings on Windows	COMMENT
LOW	tests/test-grad0.cpp	21	#define MIN(a, b) ((a) < (b) ? (a) : (b))	COMMENT
LOW	tests/test-grad0.cpp	661	struct ggml_tensor * f = ggml_sum(ctx0, ggml_sqr(ctx0, ggml_sub(ctx0, x[0], ggml_repeat_back(ctx0, x[1],	COMMENT
LOW	tests/test-grad0.cpp	1521	// const int nargs = 3;	COMMENT
LOW	tests/test-grad0.cpp	1541	// nev[2] = 1; nev[3] = 1;	COMMENT
LOW	tests/test-tokenizer-1-bpe.cpp	1	#include "llama.h"	COMMENT
LOW	tests/test-tokenizer-1-spm.cpp	1	#include "llama.h"	COMMENT
LOW	tests/test-reasoning-budget.cpp	1	#include "reasoning-budget.h"	COMMENT
LOW	tests/test-grammar-integration.cpp	1	#ifdef NDEBUG	COMMENT
LOW	tests/test-quantize-fns.cpp	1	// Unit tests for quantization specific functions - quantize, dequantize and dot product	COMMENT
LOW	tests/test-tokenizer-0.cpp	1	#include "llama.h"	COMMENT
LOW	tests/test-tokenizer-0.cpp	21	// { "\t\n" , { 1602, }, },	COMMENT
LOW	tests/test-tokenizer-0.cpp	41	// { "\n =" , { 198, 284, }, },	COMMENT
LOW	tests/test-iq4-ks-kt-decode.cpp	1	// Checks the IQ4_KS and IQ4_KT decode arithmetic against ggml's to_float.	COMMENT
LOW	tests/test-opt.cpp	1	#include "ggml.h"	COMMENT
LOW	tests/test-opt.cpp	21	#define GGML_PRINT_DEBUG(...)	COMMENT
LOW	tests/test-opt.cpp	141	// int64_t ne3[4] = {128, 256, 1, 1};	COMMENT
LOW	tests/test-opt.cpp	161	// main: original e = 126061.1953	COMMENT
LOW	tests/test-chat-peg-parser.cpp	1	#include "chat-peg-parser.h"	COMMENT
LOW	tests/test-tokenizer-0.py	21	t_start = time.time()	COMMENT
LOW	tests/test-double-float.cpp	1	// These tests may take a long time!	COMMENT
LOW	tests/test-quantize-perf.cpp	1	// Benchmark quantization specific functions on synthetic data	COMMENT
LOW	tests/test-chat-template.cpp	1	#include <string>	COMMENT
LOW	tests/test-backend-ops.cpp	1	#include <ggml.h>	COMMENT
LOW	tests/test-backend-ops.cpp	261	default: return std::to_string(pool);	COMMENT
LOW	…Guide coming over from llama.cpp and ktransformers_.md	41	# or Configure CPU Only Backend	COMMENT
LOW	…Guide coming over from llama.cpp and ktransformers_.md	61	# tl;dr; generally use -mla 2 for CPU+GPU and use -mla 3 for CPU assuming your model architecture supports MLA	COMMENT
LOW	…Guide coming over from llama.cpp and ktransformers_.md	81	-amb 512 # 512 MiB compute buffer is a good for DeepSeek-R1 671B on a single <24GB VRAM GPU	COMMENT
LOW	…Guide coming over from llama.cpp and ktransformers_.md	101	-ser 7,1 # or 6,1 or 5,1 for faster trading off quality for speed	COMMENT
LOW	…Guide coming over from llama.cpp and ktransformers_.md	121	# e.g. IQ2_K_R4 and friends for CPU tensors	COMMENT
LOW	…Guide coming over from llama.cpp and ktransformers_.md	361	--outtype bf16 \	COMMENT
LOW	…Guide coming over from llama.cpp and ktransformers_.md	421	# Q6_K / Q5_K for all non-shared expert down_proj and up_proj/gate_proj respectively	COMMENT
LOW	…a/discussions/621 - Deepseek v3_r1 poisoned prompt_.md	81		COMMENT
LOW	…a/discussions/621 - Deepseek v3_r1 poisoned prompt_.md	101	# #	COMMENT
LOW	…ub-data/discussions/434 - Quant Cookers Basic Guide.md	141	# blk.28.attn_q.weight, torch.bfloat16 --> BF16, shape = {5120, 5120}	COMMENT
LOW	…b-data/pull_requests/336 - Fix termux_android build.md	61	👤 ikawrakow commented during a code review the 2025-04-21 at 06:11:32 on `ggml/src/iqk/iqk_config.h`:<br>	COMMENT
LOW	…ests/612 - kimi-k2 convert script and chat template.md	81	# blk.0.attn_norm.weight - [ 7168, 1, 1, 1], type = f32, size = 0.027 MB	COMMENT
LOW	…ests/612 - kimi-k2 convert script and chat template.md	101	# blk.9.attn_kv_a_norm.weight - [ 512, 1, 1, 1], type = f32, size = 0.002 MB	COMMENT
LOW	… The streaming every couple of rows blocks for 5-8s.md	261	#2 gomp_barrier_wait_end (bar=0x55555f628310, state=60160) at ../../../src/libgomp/config/linux/bar.c:48	COMMENT
515 more matches not shown…

Hyper-Verbose Identifiers232 hits · 226 pts

Severity	File	Line	Snippet	Context
LOW	convert_hf_to_gguf.py	473	def get_model_part_names_from_weight_map(dir_model: Path, index_name: str) -> list[str]:	CODE
LOW	convert_hf_to_gguf.py	851	def _create_vocab_sentencepiece(self):	CODE
LOW	convert_hf_to_gguf.py	1804	def _ffn_mult_to_intermediate_size(ffn_mult: float, n_embd: int) -> int:	CODE
LOW	convert_hf_to_gguf.py	2446	def _require_target_model_dir(self) -> Path:	CODE
LOW	convert_hf_to_gguf.py	2732	def _set_vocab_tokenizer_json(self, dir_model: Path, vocab_size: int) -> None:	CODE
LOW	convert_hf_to_gguf.py	3642	def shuffle_attn_output_weight(self, data_torch):	CODE
LOW	convert_hf_to_gguf.py	7314	def _set_attention_layer_types(self):	CODE
LOW	convert_hf_to_gguf.py	7561	def meta_with_dtype_and_shape(cls, dtype: torch.dtype, shape: tuple[int, ...]) -> Tensor:	CODE
LOW⚡	gguf-py/gguf/gguf_writer.py	661	def add_leading_dense_block_count(self, length: int) -> None:	CODE
LOW⚡	gguf-py/gguf/gguf_writer.py	670	def add_expert_feed_forward_length(self, length: int) -> None:	CODE
LOW⚡	gguf-py/gguf/gguf_writer.py	673	def add_expert_shared_feed_forward_length(self, length: int) -> None:	CODE
LOW⚡	gguf-py/gguf/gguf_writer.py	676	def add_expert_chunk_feed_forward_length(self, length: int) -> None:	CODE
LOW⚡	gguf-py/gguf/gguf_writer.py	682	def add_decoder_start_token_id(self, id: int) -> None:	CODE
LOW⚡	gguf-py/gguf/gguf_writer.py	688	def add_embedding_length_per_layer_input(self, value: int) -> None:	CODE
LOW⚡	gguf-py/gguf/gguf_writer.py	697	def add_activation_sparsity_scale(self, values: Sequence[float]) -> None:	CODE
LOW⚡	gguf-py/gguf/gguf_writer.py	755	def add_attn_logit_softcapping(self, value: float) -> None:	CODE
LOW⚡	gguf-py/gguf/gguf_writer.py	758	def add_router_logit_softcapping(self, value: float) -> None:	CODE
LOW⚡	gguf-py/gguf/gguf_writer.py	761	def add_final_logit_softcapping(self, value: float) -> None:	CODE
LOW⚡	gguf-py/gguf/gguf_writer.py	812	def add_attention_indexer_head_count(self, count: int) -> None:	CODE
LOW⚡	gguf-py/gguf/gguf_writer.py	815	def add_attention_indexer_key_length(self, length: int) -> None:	CODE
LOW⚡	gguf-py/gguf/gguf_writer.py	818	def add_attention_indexer_top_k(self, top_k: int) -> None:	CODE
LOW⚡	gguf-py/gguf/gguf_writer.py	821	def add_attention_indexer_block_size(self, size: int) -> None:	CODE
LOW⚡	gguf-py/gguf/gguf_writer.py	824	def add_attention_indexer_local_blocks(self, blocks: int) -> None:	CODE
LOW⚡	gguf-py/gguf/gguf_writer.py	830	def add_rescale_every_n_layers(self, count: int) -> None:	CODE
LOW⚡	gguf-py/gguf/gguf_writer.py	881	def add_value_residual_mix_lora_rank(self, length: int) -> None:	CODE
LOW⚡	gguf-py/gguf/gguf_writer.py	887	def add_relative_attn_buckets_count(self, value: int) -> None:	CODE
LOW⚡	gguf-py/gguf/gguf_writer.py	896	def add_attention_value_scale(self, value: float) -> None:	CODE
LOW⚡	gguf-py/gguf/gguf_writer.py	902	def add_attn_temperature_length(self, value: int) -> None:	CODE
LOW⚡	gguf-py/gguf/gguf_writer.py	932	def add_rope_scaling_attn_factors(self, value: float) -> None:	CODE
LOW⚡	gguf-py/gguf/gguf_writer.py	935	def add_rope_scaling_orig_ctx_len(self, value: int) -> None:	CODE
LOW⚡	gguf-py/gguf/gguf_writer.py	938	def add_rope_scaling_finetuned(self, value: bool) -> None:	CODE
LOW⚡	gguf-py/gguf/gguf_writer.py	941	def add_rope_scaling_yarn_log_mul(self, value: float) -> None:	CODE
LOW⚡	gguf-py/gguf/gguf_writer.py	944	def add_rope_scaling_yarn_ext_factor(self, value: float) -> None:	CODE
LOW⚡	gguf-py/gguf/gguf_writer.py	947	def add_rope_scaling_yarn_attn_factor(self, value: float) -> None:	CODE
LOW⚡	gguf-py/gguf/gguf_writer.py	950	def add_rope_scaling_yarn_beta_fast(self, value: float) -> None:	CODE
LOW⚡	gguf-py/gguf/gguf_writer.py	953	def add_rope_scaling_yarn_beta_slow(self, value: float) -> None:	CODE
LOW⚡	gguf-py/gguf/gguf_writer.py	986	def add_activation_situ_linear_beta(self, value: float) -> None:	CODE
LOW⚡	gguf-py/gguf/gguf_writer.py	989	def add_hyper_connection_count(self, value: int) -> None:	CODE
LOW⚡	gguf-py/gguf/gguf_writer.py	992	def add_hyper_connection_epsilon(self, value: float) -> None:	CODE
LOW⚡	gguf-py/gguf/gguf_writer.py	995	def add_hyper_connection_magnitude(self, value: float) -> None:	CODE
LOW⚡	gguf-py/gguf/gguf_writer.py	1090	def add_classifier_output_labels(self, labels: Sequence[str]) -> None:	CODE
LOW⚡	gguf-py/gguf/gguf_writer.py	1095	def add_clip_has_vision_encoder(self, value: bool) -> None:	CODE
LOW⚡	gguf-py/gguf/gguf_writer.py	1098	def add_clip_has_audio_encoder(self, value: bool) -> None:	CODE
LOW⚡	gguf-py/gguf/gguf_writer.py	1104	def add_vision_projection_dim(self, value: int) -> None:	CODE
LOW⚡	gguf-py/gguf/gguf_writer.py	1110	def add_vision_embedding_length(self, value: int) -> None:	CODE
LOW⚡	gguf-py/gguf/gguf_writer.py	1113	def add_vision_feed_forward_length(self, value: int) -> None:	CODE
LOW⚡	gguf-py/gguf/gguf_writer.py	1122	def add_vision_attention_layernorm_eps(self, value: float) -> None:	CODE
LOW⚡	gguf-py/gguf/gguf_writer.py	1128	def add_vision_preproc_image_size(self, value: int) -> None:	CODE
LOW⚡	gguf-py/gguf/gguf_writer.py	1137	def add_vision_spatial_merge_size(self, value: int) -> None:	CODE
LOW⚡	gguf-py/gguf/gguf_writer.py	1146	def add_vision_projector_scale_factor(self, value: int) -> None:	CODE
LOW⚡	gguf-py/gguf/gguf_writer.py	1152	def add_vision_is_deepstack_layers(self, layers: Sequence[bool]) -> None:	CODE
LOW⚡	gguf-py/gguf/gguf_writer.py	1160	def add_audio_embedding_length(self, value: int) -> None:	CODE
LOW⚡	gguf-py/gguf/gguf_writer.py	1163	def add_audio_feed_forward_length(self, value: int) -> None:	CODE
LOW⚡	gguf-py/gguf/gguf_writer.py	1172	def add_audio_attention_layernorm_eps(self, value: float) -> None:	CODE
LOW	gguf-py/gguf/gguf_writer.py	112	def get_total_parameter_count(self) -> tuple[int, int, int, int]:	CODE
LOW	gguf-py/gguf/gguf_writer.py	571	def add_base_model_organization(self, source_id: int, organization: str) -> None:	CODE
LOW	gguf-py/gguf/gguf_writer.py	574	def add_base_model_description(self, source_id: int, description: str) -> None:	CODE
LOW	gguf-py/gguf/gguf_writer.py	643	def add_posnet_embedding_length(self, length: int) -> None:	CODE
LOW	gguf-py/gguf/gguf_writer.py	649	def add_convnext_embedding_length(self, length: int) -> None:	CODE
LOW	gguf-py/gguf/gguf_writer.py	739	def add_sliding_window_pattern(self, value: Sequence[bool]) -> None:	CODE
172 more matches not shown…

Cross-File Repetition25 hits · 125 pts

Severity	File	Snippet	Context
HIGH	gguf-py/gguf/vocab.py	returns list of utf-8 byte and a corresponding list of unicode strings. the reversible bpe codes work on unicode strings	STRING
HIGH	…legacy-models/glmedge-convert-image-encoder-to-gguf.py	returns list of utf-8 byte and a corresponding list of unicode strings. the reversible bpe codes work on unicode strings	STRING
HIGH	…es/mtmd/legacy-models/convert_image_encoder_to_gguf.py	returns list of utf-8 byte and a corresponding list of unicode strings. the reversible bpe codes work on unicode strings	STRING
HIGH	…egacy-models/minicpmv-convert-image-encoder-to-gguf.py	returns list of utf-8 byte and a corresponding list of unicode strings. the reversible bpe codes work on unicode strings	STRING
HIGH	tests/test-grammar-llguidance.cpp	({ "type": "integer", "minimum": 5, "maximum": 30 })	STRING
HIGH	tests/test-grammar-integration.cpp	({ "type": "integer", "minimum": 5, "maximum": 30 })	STRING
HIGH	tests/test-json-schema-to-grammar.cpp	({ "type": "integer", "minimum": 5, "maximum": 30 })	STRING
HIGH	tests/test-grammar-llguidance.cpp	({ "type": "integer", "minimum": -123, "maximum": 42 })	STRING
HIGH	tests/test-grammar-integration.cpp	({ "type": "integer", "minimum": -123, "maximum": 42 })	STRING
HIGH	tests/test-json-schema-to-grammar.cpp	({ "type": "integer", "minimum": -123, "maximum": 42 })	STRING
HIGH	tests/test-grammar-llguidance.cpp	({ "items": [ { "format": "date" }, { "format": "uuid" }, { "format": "time" }, { "format": "date-time" } ] })	STRING
HIGH	tests/test-grammar-integration.cpp	({ "items": [ { "format": "date" }, { "format": "uuid" }, { "format": "time" }, { "format": "date-time" } ] })	STRING
HIGH	tests/test-json-schema-to-grammar.cpp	({ "items": [ { "format": "date" }, { "format": "uuid" }, { "format": "time" }, { "format": "date-time" } ] })	STRING
HIGH	tests/test-grammar-llguidance.cpp	({ "type": "string", "minlength": 1, "maxlength": 4 })	STRING
HIGH	tests/test-grammar-integration.cpp	({ "type": "string", "minlength": 1, "maxlength": 4 })	STRING
HIGH	tests/test-json-schema-to-grammar.cpp	({ "type": "string", "minlength": 1, "maxlength": 4 })	STRING
HIGH	tests/test-grammar-llguidance.cpp	({ "enum": ["red", "amber", "green", null, 42, ["foo"]] })	STRING
HIGH	tests/test-grammar-integration.cpp	({ "enum": ["red", "amber", "green", null, 42, ["foo"]] })	STRING
HIGH	tests/test-json-schema-to-grammar.cpp	({ "enum": ["red", "amber", "green", null, 42, ["foo"]] })	STRING
HIGH	tests/test-grammar-llguidance.cpp	({ "items": { "type": ["number", "integer"] }, "minitems": 3, "maxitems": 5 })	STRING
HIGH	tests/test-grammar-integration.cpp	({ "items": { "type": ["number", "integer"] }, "minitems": 3, "maxitems": 5 })	STRING
HIGH	tests/test-json-schema-to-grammar.cpp	({ "items": { "type": ["number", "integer"] }, "minitems": 3, "maxitems": 5 })	STRING
HIGH	tests/test-grammar-llguidance.cpp	({ "properties": { "b": {"type": "string"}, "a": {"type": "string"}, "d": {"type": "string"}, "c": {"type": "string"} },	STRING
HIGH	tests/test-grammar-integration.cpp	({ "properties": { "b": {"type": "string"}, "a": {"type": "string"}, "d": {"type": "string"}, "c": {"type": "string"} },	STRING
HIGH	tests/test-json-schema-to-grammar.cpp	({ "properties": { "b": {"type": "string"}, "a": {"type": "string"}, "d": {"type": "string"}, "c": {"type": "string"} },	STRING

Structural Annotation Overuse68 hits · 123 pts

Severity	File	Line	Snippet	Context
LOW	include/llama.h	1377	/// NOTE: This function does not use a jinja parser. It only support a pre-defined list of template. See more: https	COMMENT
LOW⚡	tests/test-function-calls.cpp	1037	// Step 2: Parse incremental message (what server does)	COMMENT
LOW⚡	tests/test-function-calls.cpp	1043	// Step 3: Compute diffs (what server streaming does)	COMMENT
LOW⚡	tests/test-function-calls.cpp	1048	// Step 4: Generate streaming response (critical part that was missing)	COMMENT
LOW⚡	tests/test-function-calls.cpp	1295	// Step 1: During streaming, partial content should not generate diffs	COMMENT
LOW⚡	tests/test-function-calls.cpp	1304	// Step 2: Only complete tool call should generate tool call diffs	COMMENT
LOW⚡	tests/test-function-calls.cpp	1345	// Step 1: Extract content like the original server does	COMMENT
LOW⚡	tests/test-function-calls.cpp	1349	// Step 2: Test our tool_call_mode fix (force content="" when ctx_server exists)	COMMENT
LOW⚡	tests/test-function-calls.cpp	1356	// Step 3: Simulate slot processing	COMMENT
LOW⚡	tests/test-function-calls.cpp	1365	// Step 4: Test our incremental parsing fix	COMMENT
LOW⚡	tests/test-function-calls.cpp	1373	// Step 5: Test our diff computation fix	COMMENT
LOW⚡	tests/test-function-calls.cpp	1378	// Step 6: Test our early return logic (diffs.empty() → return empty chunks)	COMMENT
LOW⚡	tests/test-function-calls.cpp	1382	// Step 7: Test fallback content logic	COMMENT
LOW⚡	tests/test-function-calls.cpp	1387	// Step 8: Validate our complete fix	COMMENT
LOW⚡	tests/test-function-calls.cpp	1874	// Step 1: Simulate the exact content from logs	COMMENT
LOW⚡	tests/test-function-calls.cpp	1880	// Step 2: Parse using current implementation (partial=true, then partial=false)	COMMENT
LOW⚡	tests/test-function-calls.cpp	1922	// Step 4: Compute diffs (this should be 0 if states are identical)	COMMENT
LOW⚡	tests/test-function-calls.cpp	1928	// Step 5: Check for contamination in diffs (if any)	COMMENT
LOW⚡	tests/test-function-calls.cpp	1943	// Step 6: CRITICAL TEST - Check raw content vs processed content disparity	COMMENT
LOW⚡	tests/test-function-calls.cpp	1951	// Step 7: REPRODUCTION CHECK - The exact issue from logs	COMMENT
LOW	tests/test-function-calls.cpp	1026	// Step 1: Simulate what format_partial_response_oaicompat() should do	COMMENT
LOW	tests/test-function-calls.cpp	1094	// Step 5: Validate the output format	COMMENT
LOW	tests/test-function-calls.cpp	1899	// Step 3: Test differential streaming scenario from logs	COMMENT
LOW	tests/test-function-calls.cpp	1985	// Step 8: Test the exact format_partial_response_oaicompat scenario	COMMENT
LOW	tests/test-function-calls.cpp	2017	// Step 9: CRITICAL TEST - Check for content duplication	COMMENT
LOW	tests/test-function-calls.cpp	2055	// Step 10: Additional duplication scenarios	COMMENT
LOW⚡	docs/development/debugging-tests.md	41	#### Step 1: Reset and Setup folder context	COMMENT
LOW⚡	docs/development/debugging-tests.md	49	#### Step 2: Setup Build Environment and Compile Test Binaries	COMMENT
LOW⚡	docs/development/debugging-tests.md	58	#### Step 3: Find all tests available that match REGEX	COMMENT
LOW	docs/development/debugging-tests.md	86	#### Step 4: Identify Test Command for Debugging	COMMENT
LOW	docs/development/debugging-tests.md	92	#### Step 5: Run GDB on test command	COMMENT
LOW⚡	…quests/386 - FlashMLA-3 for DeepSeek models on CUDA.md	1057	### Step 1: Add the two equations to eliminate $ y $	COMMENT
LOW⚡	…quests/386 - FlashMLA-3 for DeepSeek models on CUDA.md	1065	### Step 2: Solve for $ x $	COMMENT
LOW⚡	…quests/386 - FlashMLA-3 for DeepSeek models on CUDA.md	1070	### Step 3: Substitute $ x = \frac{5}{2} $ into one of the original equations to find $ y $	COMMENT
LOW⚡	…ix CUDA DeepSeek FlashMLA-3 with quantized KV cache.md	610	### Step 1: Add the two equations	COMMENT
LOW⚡	…ix CUDA DeepSeek FlashMLA-3 with quantized KV cache.md	616	### Step 2: Solve for $ x $	COMMENT
LOW⚡	…ix CUDA DeepSeek FlashMLA-3 with quantized KV cache.md	622	### Step 3: Substitute $ x = 2.5 $ into one of the original equations to find $ y $	COMMENT
LOW⚡	…ix CUDA DeepSeek FlashMLA-3 with quantized KV cache.md	701	### Step 1: Analyze the Lengths	COMMENT
LOW⚡	…ix CUDA DeepSeek FlashMLA-3 with quantized KV cache.md	707	### Step 2: Identify Letter Pairs	COMMENT
LOW⚡	…ix CUDA DeepSeek FlashMLA-3 with quantized KV cache.md	716	### Step 3: Match Letters	COMMENT
LOW	…ix CUDA DeepSeek FlashMLA-3 with quantized KV cache.md	735	### Step 4: Formulate the Encoding Rule	COMMENT
LOW	…erver/webui_llamacpp/src/lib/utils/latex-protection.ts	151	// Step 0: Temporarily remove blockquote markers (>) to process LaTeX correctly	COMMENT
LOW⚡	…erver/webui_llamacpp/src/lib/utils/latex-protection.ts	165	// Step 1: Protect code blocks	COMMENT
LOW⚡	…erver/webui_llamacpp/src/lib/utils/latex-protection.ts	174	// Step 2: Protect existing LaTeX expressions	COMMENT
LOW⚡	…erver/webui_llamacpp/src/lib/utils/latex-protection.ts	210	// Step 3: Escape standalone $ before digits (currency like $5 → \$5)	COMMENT
LOW⚡	…erver/webui_llamacpp/src/lib/utils/latex-protection.ts	214	// Step 4: Restore protected LaTeX expressions (they are valid)	COMMENT
LOW⚡	…erver/webui_llamacpp/src/lib/utils/latex-protection.ts	229	// Step 5: Apply additional escaping functions (brackets and mhchem)	COMMENT
LOW⚡	…erver/webui_llamacpp/src/lib/utils/latex-protection.ts	237	// Step 6: Convert remaining $...$ → $...$, \[...\] → $$...$$	COMMENT
LOW⚡	…erver/webui_llamacpp/src/lib/utils/latex-protection.ts	253	// Step 7: Restore code blocks	COMMENT
LOW⚡	…erver/webui_llamacpp/src/lib/utils/latex-protection.ts	259	// Step 8: Restore blockquote markers	COMMENT
LOW⚡	…amples/server/webui/src/components/MarkdownDisplay.tsx	233	// Step 1: Protect code blocks	COMMENT
LOW⚡	…amples/server/webui/src/components/MarkdownDisplay.tsx	240	// Step 2: Protect existing LaTeX expressions	COMMENT
LOW⚡	…amples/server/webui/src/components/MarkdownDisplay.tsx	266	// Step 3: Escape dollar signs that are likely currency indicators.	COMMENT
LOW⚡	…amples/server/webui/src/components/MarkdownDisplay.tsx	270	// Step 4: Restore LaTeX expressions	COMMENT
LOW⚡	…amples/server/webui/src/components/MarkdownDisplay.tsx	276	// Step 5: Restore code blocks	COMMENT
LOW⚡	…amples/server/webui/src/components/MarkdownDisplay.tsx	282	// Step 6: Apply additional escaping functions	COMMENT
LOW	scripts/check-requirements.sh	8	# WARNING: This is quite IO intensive, because a fresh venv is set up for every	COMMENT
LOW	scripts/debug-test.sh	62	# Step 0: Check the args	COMMENT
LOW	scripts/debug-test.sh	95	# Step 1: Reset and Setup folder context	COMMENT
LOW⚡	scripts/debug-test.sh	109	# Step 2: Setup Build Environment and Compile Test Binaries	COMMENT
8 more matches not shown…

Verbosity Indicators51 hits · 88 pts

Severity	File	Line	Snippet	Context
LOW⚡	tests/test-function-calls.cpp	1037	// Step 2: Parse incremental message (what server does)	COMMENT
LOW⚡	tests/test-function-calls.cpp	1043	// Step 3: Compute diffs (what server streaming does)	COMMENT
LOW⚡	tests/test-function-calls.cpp	1048	// Step 4: Generate streaming response (critical part that was missing)	COMMENT
LOW⚡	tests/test-function-calls.cpp	1295	// Step 1: During streaming, partial content should not generate diffs	COMMENT
LOW⚡	tests/test-function-calls.cpp	1304	// Step 2: Only complete tool call should generate tool call diffs	COMMENT
LOW⚡	tests/test-function-calls.cpp	1345	// Step 1: Extract content like the original server does	COMMENT
LOW⚡	tests/test-function-calls.cpp	1349	// Step 2: Test our tool_call_mode fix (force content="" when ctx_server exists)	COMMENT
LOW⚡	tests/test-function-calls.cpp	1356	// Step 3: Simulate slot processing	COMMENT
LOW⚡	tests/test-function-calls.cpp	1365	// Step 4: Test our incremental parsing fix	COMMENT
LOW⚡	tests/test-function-calls.cpp	1373	// Step 5: Test our diff computation fix	COMMENT
LOW⚡	tests/test-function-calls.cpp	1378	// Step 6: Test our early return logic (diffs.empty() → return empty chunks)	COMMENT
LOW⚡	tests/test-function-calls.cpp	1382	// Step 7: Test fallback content logic	COMMENT
LOW⚡	tests/test-function-calls.cpp	1387	// Step 8: Validate our complete fix	COMMENT
LOW⚡	tests/test-function-calls.cpp	1874	// Step 1: Simulate the exact content from logs	COMMENT
LOW⚡	tests/test-function-calls.cpp	1880	// Step 2: Parse using current implementation (partial=true, then partial=false)	COMMENT
LOW⚡	tests/test-function-calls.cpp	1922	// Step 4: Compute diffs (this should be 0 if states are identical)	COMMENT
LOW⚡	tests/test-function-calls.cpp	1928	// Step 5: Check for contamination in diffs (if any)	COMMENT
LOW⚡	tests/test-function-calls.cpp	1943	// Step 6: CRITICAL TEST - Check raw content vs processed content disparity	COMMENT
LOW⚡	tests/test-function-calls.cpp	1951	// Step 7: REPRODUCTION CHECK - The exact issue from logs	COMMENT
LOW	tests/test-function-calls.cpp	1026	// Step 1: Simulate what format_partial_response_oaicompat() should do	COMMENT
LOW	tests/test-function-calls.cpp	1094	// Step 5: Validate the output format	COMMENT
LOW	tests/test-function-calls.cpp	1899	// Step 3: Test differential streaming scenario from logs	COMMENT
LOW	tests/test-function-calls.cpp	1985	// Step 8: Test the exact format_partial_response_oaicompat scenario	COMMENT
LOW	tests/test-function-calls.cpp	2017	// Step 9: CRITICAL TEST - Check for content duplication	COMMENT
LOW	tests/test-function-calls.cpp	2055	// Step 10: Additional duplication scenarios	COMMENT
LOW	common/common.cpp	3269	// now we need to check if there is anything after this argv	COMMENT
LOW	…erver/webui_llamacpp/src/lib/utils/latex-protection.ts	151	// Step 0: Temporarily remove blockquote markers (>) to process LaTeX correctly	COMMENT
LOW⚡	…erver/webui_llamacpp/src/lib/utils/latex-protection.ts	165	// Step 1: Protect code blocks	COMMENT
LOW⚡	…erver/webui_llamacpp/src/lib/utils/latex-protection.ts	174	// Step 2: Protect existing LaTeX expressions	COMMENT
LOW⚡	…erver/webui_llamacpp/src/lib/utils/latex-protection.ts	210	// Step 3: Escape standalone $ before digits (currency like $5 → \$5)	COMMENT
LOW⚡	…erver/webui_llamacpp/src/lib/utils/latex-protection.ts	214	// Step 4: Restore protected LaTeX expressions (they are valid)	COMMENT
LOW⚡	…erver/webui_llamacpp/src/lib/utils/latex-protection.ts	229	// Step 5: Apply additional escaping functions (brackets and mhchem)	COMMENT
LOW⚡	…erver/webui_llamacpp/src/lib/utils/latex-protection.ts	237	// Step 6: Convert remaining $...$ → $...$, \[...\] → $$...$$	COMMENT
LOW⚡	…erver/webui_llamacpp/src/lib/utils/latex-protection.ts	253	// Step 7: Restore code blocks	COMMENT
LOW⚡	…erver/webui_llamacpp/src/lib/utils/latex-protection.ts	259	// Step 8: Restore blockquote markers	COMMENT
LOW⚡	…amples/server/webui/src/components/MarkdownDisplay.tsx	233	// Step 1: Protect code blocks	COMMENT
LOW⚡	…amples/server/webui/src/components/MarkdownDisplay.tsx	240	// Step 2: Protect existing LaTeX expressions	COMMENT
LOW⚡	…amples/server/webui/src/components/MarkdownDisplay.tsx	266	// Step 3: Escape dollar signs that are likely currency indicators.	COMMENT
LOW⚡	…amples/server/webui/src/components/MarkdownDisplay.tsx	270	// Step 4: Restore LaTeX expressions	COMMENT
LOW⚡	…amples/server/webui/src/components/MarkdownDisplay.tsx	276	// Step 5: Restore code blocks	COMMENT
LOW⚡	…amples/server/webui/src/components/MarkdownDisplay.tsx	282	// Step 6: Apply additional escaping functions	COMMENT
LOW	scripts/debug-test.sh	62	# Step 0: Check the args	COMMENT
LOW	scripts/debug-test.sh	95	# Step 1: Reset and Setup folder context	COMMENT
LOW⚡	scripts/debug-test.sh	109	# Step 2: Setup Build Environment and Compile Test Binaries	COMMENT
LOW⚡	scripts/debug-test.sh	119	# Step 3: Find all tests available that matches REGEX	COMMENT
LOW	scripts/debug-test.sh	135	# Step 4: Identify Test Command for Debugging	COMMENT
LOW	scripts/debug-test.sh	172	# Step 5: Execute or GDB Debug	COMMENT
LOW	src/llama-sampling.cpp	830	// Step 1: Look for restart sequences to limit the maximum repetition length.	COMMENT
LOW	src/llama-sampling.cpp	891	// Step 2: Iterate in reverse over the last N tokens of the context, using the "Z-algorithm" (in	COMMENT
LOW	src/llama-sampling.cpp	955	// Step 3: Iterate over dry_repeat_count and last_tokens, examining the maximum repeat length	COMMENT
LOW	src/llama-sampling.cpp	982	// Step 4: Apply logit penalties based on the maximum repeat length for relevant tokens.	COMMENT

Deep Nesting98 hits · 74 pts

Severity	File	Line	Context
LOW	convert_lora_to_gguf.py	56	CODE
LOW	convert_lora_to_gguf.py	186	CODE
LOW	convert_imatrix_gguf_to_dat.py	87	CODE
LOW	convert_hf_to_gguf.py	7655	CODE
LOW	convert_hf_to_gguf.py	139	CODE
LOW	convert_hf_to_gguf.py	273	CODE
LOW	convert_hf_to_gguf.py	544	CODE
LOW	convert_hf_to_gguf.py	851	CODE
LOW	convert_hf_to_gguf.py	1283	CODE
LOW	convert_hf_to_gguf.py	1493	CODE
LOW	convert_hf_to_gguf.py	1695	CODE
LOW	convert_hf_to_gguf.py	1740	CODE
LOW	convert_hf_to_gguf.py	1816	CODE
LOW	convert_hf_to_gguf.py	1944	CODE
LOW	convert_hf_to_gguf.py	2000	CODE
LOW	convert_hf_to_gguf.py	2089	CODE
LOW	convert_hf_to_gguf.py	2325	CODE
LOW	convert_hf_to_gguf.py	2536	CODE
LOW	convert_hf_to_gguf.py	3220	CODE
LOW	convert_hf_to_gguf.py	3327	CODE
LOW	convert_hf_to_gguf.py	3456	CODE
LOW	convert_hf_to_gguf.py	3703	CODE
LOW	convert_hf_to_gguf.py	3989	CODE
LOW	convert_hf_to_gguf.py	4777	CODE
LOW	convert_hf_to_gguf.py	4869	CODE
LOW	convert_hf_to_gguf.py	4929	CODE
LOW	convert_hf_to_gguf.py	5028	CODE
LOW	convert_hf_to_gguf.py	5139	CODE
LOW	convert_hf_to_gguf.py	5223	CODE
LOW	convert_hf_to_gguf.py	5759	CODE
LOW	convert_hf_to_gguf.py	5865	CODE
LOW	convert_hf_to_gguf.py	6017	CODE
LOW	convert_hf_to_gguf.py	6087	CODE
LOW	convert_hf_to_gguf.py	6227	CODE
LOW	convert_hf_to_gguf.py	6711	CODE
LOW	convert_hf_to_gguf.py	6822	CODE
LOW	convert_hf_to_gguf.py	6886	CODE
LOW	convert_hf_to_gguf.py	6976	CODE
LOW	convert_hf_to_gguf.py	7115	CODE
LOW	convert_llama_ggml_to_gguf.py	204	CODE
LOW	convert_llama_ggml_to_gguf.py	287	CODE
LOW	gguf-py/gguf/gguf_writer.py	112	CODE
LOW	gguf-py/gguf/gguf_writer.py	326	CODE
LOW	gguf-py/gguf/gguf_writer.py	425	CODE
LOW	gguf-py/gguf/gguf_writer.py	1055	CODE
LOW	gguf-py/gguf/gguf_writer.py	1204	CODE
LOW	gguf-py/gguf/vocab.py	141	CODE
LOW	gguf-py/gguf/metadata.py	155	CODE
LOW	gguf-py/gguf/metadata.py	280	CODE
LOW	gguf-py/gguf/gguf_reader.py	56	CODE
LOW	gguf-py/gguf/gguf_reader.py	314	CODE
LOW	gguf-py/gguf/constants.py	2490	CODE
LOW	gguf-py/gguf/quants.py	50	CODE
LOW	gguf-py/tests/test_quants.py	142	CODE
LOW	gguf-py/tests/test_quants.py	85	CODE
LOW	gguf-py/scripts/gguf_convert_endian.py	22	CODE
LOW	gguf-py/scripts/gguf_new_metadata.py	136	CODE
LOW	gguf-py/scripts/gguf_dump.py	34	CODE
LOW	gguf-py/scripts/gguf_dump.py	153	CODE
LOW	gguf-py/scripts/gguf_dump.py	228	CODE
38 more matches not shown…

Excessive Try-Catch Wrapping36 hits · 59 pts

Severity	File	Line	Snippet	Context
MEDIUM	convert_hf_to_gguf.py	1124	def set_vocab(self):	CODE
LOW	convert_hf_to_gguf.py	1127	except Exception:	CODE
LOW	convert_hf_to_gguf_update.py	157	except Exception as e:	CODE
MEDIUM	…a/discussions/621 - Deepseek v3_r1 poisoned prompt_.md	159	print(f"Error: tensor_downloader.sh not found or not executable at {tensor_downloader}", file=sys.stderr)	CODE
LOW	…a/discussions/621 - Deepseek v3_r1 poisoned prompt_.md	220	except Exception as e:	CODE
LOW	…a/discussions/621 - Deepseek v3_r1 poisoned prompt_.md	312	except Exception:	CODE
LOW	…a/discussions/621 - Deepseek v3_r1 poisoned prompt_.md	330	except Exception:	CODE
MEDIUM	…a/discussions/621 - Deepseek v3_r1 poisoned prompt_.md	397	print(f"Error: qtype '{qtype_arg}' not found in CSV.")	CODE
LOW⚡	…a/discussions/621 - Deepseek v3_r1 poisoned prompt_.md	782	except Exception:	CODE
LOW⚡	…a/discussions/621 - Deepseek v3_r1 poisoned prompt_.md	790	except Exception:	CODE
MEDIUM⚡	…a/discussions/621 - Deepseek v3_r1 poisoned prompt_.md	796	print("Error: CSV must have 'QTYPE' as first column.")	CODE
MEDIUM	…a/discussions/621 - Deepseek v3_r1 poisoned prompt_.md	864	print(f"Error: could not parse numeric value for tensor {name!r}: {raw!r}")	CODE
LOW	…a/discussions/621 - Deepseek v3_r1 poisoned prompt_.md	1165	except Exception:	CODE
MEDIUM	examples/pydantic_models_to_grammar_examples.py	66	print(f"Error: unknown tool {json_data['function']}")	CODE
MEDIUM	examples/pydantic_models_to_grammar_examples.py	142	print(f"Error: unknown tool {json_data['function']}")	CODE
MEDIUM	examples/pydantic_models_to_grammar_examples.py	285	print(f"Error: unknown tool {call['function']}")	CODE
MEDIUM	examples/pydantic_models_to_grammar.py	1001	print(f"An error occurred while saving the grammar file: {e}")	CODE
MEDIUM	examples/pydantic_models_to_grammar.py	1008	print(f"An error occurred while saving the documentation file: {e}")	CODE
LOW	examples/server/bench/bench.py	58	except Exception:	CODE
LOW	examples/server/bench/bench.py	84	except Exception:	CODE
LOW	examples/server/tests/features/environment.py	51	except Exception:	CODE
MEDIUM	examples/server/tests/features/environment.py	23	def after_scenario(context, scenario):	CODE
MEDIUM	…es/llama.swiftui/llama.swiftui/UI/DownloadButton.swift	40	print("Error: \(error.localizedDescription)")	CODE
MEDIUM	…es/llama.swiftui/llama.swiftui/UI/DownloadButton.swift	61	print("Error: \(err.localizedDescription)")	CODE
MEDIUM	…es/llama.swiftui/llama.swiftui/UI/DownloadButton.swift	95	print("Error: \(err.localizedDescription)")	CODE
MEDIUM	…mples/llama.swiftui/llama.swiftui/UI/InputButton.swift	44	print("Error: \(error.localizedDescription)")	CODE
MEDIUM	…mples/llama.swiftui/llama.swiftui/UI/InputButton.swift	65	print("Error: \(err.localizedDescription)")	CODE
MEDIUM	…mples/llama.swiftui/llama.swiftui/UI/InputButton.swift	111	print("Error: \(err.localizedDescription)")	CODE
MEDIUM	…/llama.swiftui/llama.swiftui/UI/LoadCustomButton.swift	34	print("Error: \(err.localizedDescription)")	CODE
MEDIUM	…mples/llama.swiftui/llama.swiftui/UI/ContentView.swift	87	print("Error deleting file: \(error)")	CODE
MEDIUM	…es/llama.swiftui/llama.swiftui/Models/LlamaState.swift	39	print("Error loading models from disk: \(error)")	CODE
LOW	scripts/fetch_server_test_models.py	35	except Exception as e:	CODE
MEDIUM	scripts/fetch_server_test_models.py	31	def collect_hf_model_test_parameters(test_file) -> Generator[HuggingFaceModel, N	CODE
LOW	scripts/tool_bench.py	103	except Exception as e:	CODE
LOW	scripts/tool_bench.py	283	except Exception as e:	CODE
LOW	scripts/repack_ternary_to_prism.py	21	except Exception:	CODE

Unused Imports47 hits · 46 pts

Severity	File	Line	Context
LOW	convert_lora_to_gguf.py	4	CODE
LOW	convert_imatrix_gguf_to_dat.py	1	CODE
LOW	convert_hf_to_gguf.py	4	CODE
LOW	convert_llama_ggml_to_gguf.py	2	CODE
LOW	gguf-py/gguf/gguf_writer.py	1	CODE
LOW	gguf-py/gguf/vocab.py	1	CODE
LOW	gguf-py/gguf/metadata.py	1	CODE
LOW	gguf-py/gguf/gguf_reader.py	5	CODE
LOW	gguf-py/gguf/tensor_mapping.py	1	CODE
LOW	gguf-py/gguf/utility.py	1	CODE
LOW	gguf-py/gguf/constants.py	1	CODE
LOW	gguf-py/gguf/__init__.py	1	CODE
LOW	gguf-py/gguf/__init__.py	2	CODE
LOW	gguf-py/gguf/__init__.py	3	CODE
LOW	gguf-py/gguf/__init__.py	4	CODE
LOW	gguf-py/gguf/__init__.py	5	CODE
LOW	gguf-py/gguf/__init__.py	6	CODE
LOW	gguf-py/gguf/__init__.py	7	CODE
LOW	gguf-py/gguf/__init__.py	8	CODE
LOW	gguf-py/gguf/__init__.py	9	CODE
LOW	gguf-py/gguf/lazy.py	1	CODE
LOW	gguf-py/gguf/quants.py	1	CODE
LOW	gguf-py/tests/__init__.py	1	CODE
LOW	gguf-py/tests/test_quants.py	7	CODE
LOW	gguf-py/scripts/gguf_convert_endian.py	2	CODE
LOW	gguf-py/scripts/gguf_new_metadata.py	2	CODE
LOW	gguf-py/scripts/__init__.py	3	CODE
LOW	gguf-py/scripts/__init__.py	4	CODE
LOW	gguf-py/scripts/__init__.py	5	CODE
LOW	gguf-py/scripts/__init__.py	6	CODE
LOW	gguf-py/scripts/gguf_dump.py	2	CODE
LOW	gguf-py/scripts/gguf_dump.py	12	CODE
LOW	gguf-py/scripts/gguf_hash.py	2	CODE
LOW	tests/test-tokenizer-random.py	9	CODE
LOW	examples/json_schema_to_grammar.py	2	CODE
LOW	examples/pydantic_models_to_grammar_examples.py	5	CODE
LOW	examples/convert_legacy_llama.py	2	CODE
LOW	examples/pydantic_models_to_grammar.py	1	CODE
LOW	…gacy-models/minimaxm3-convert-image-encoder-to-gguf.py	4	CODE
LOW	…gacy-models/minimaxm3-convert-image-encoder-to-gguf.py	5	CODE
LOW	…legacy-models/glmedge-convert-image-encoder-to-gguf.py	8	CODE
LOW	…es/mtmd/legacy-models/convert_image_encoder_to_gguf.py	8	CODE
LOW	…egacy-models/minicpmv-convert-image-encoder-to-gguf.py	414	CODE
LOW	examples/server/bench/bench.py	1	CODE
LOW	scripts/gen-unicode-data.py	1	CODE
LOW	scripts/gen-unicode-script-data.py	2	CODE
LOW	scripts/repack_ternary_to_prism.py	15	CODE

Self-Referential Comments13 hits · 44 pts

Severity	File	Line	Snippet	Context
MEDIUM	ci/run.sh	742	# Create a fresh python3 venv and enter it	COMMENT
MEDIUM	gguf-py/gguf/metadata.py	52	# Create a new Metadata instance	COMMENT
MEDIUM	examples/ts-type-to-grammar.sh	10	# Create a temporary directory	COMMENT
MEDIUM	examples/pydantic_models_to_grammar.py	186	# Define the rule identifier based on max_digit and min_digit	COMMENT
MEDIUM	examples/pydantic_models_to_grammar.py	241	# Define the integer part rule	COMMENT
MEDIUM	examples/pydantic_models_to_grammar.py	248	# Define the fractional part rule based on precision constraints	COMMENT
MEDIUM	examples/pydantic_models_to_grammar.py	264	# Define the float rule	COMMENT
MEDIUM	examples/pydantic_models_to_grammar.py	370	# Defining the union grammar rule separately	COMMENT
MEDIUM	examples/pydantic_models_to_grammar.py	1179	# Creating the dynamic model	COMMENT
MEDIUM	examples/server/chat-llama2.sh	57	# Create a temporary file to hold the Python output	COMMENT
MEDIUM⚡	scripts/verify-checksum-models.py	25	# Define the path to the llama directory (parent folder of script directory)	COMMENT
MEDIUM⚡	scripts/verify-checksum-models.py	28	# Define the file with the list of hashes and filenames	COMMENT
MEDIUM⚡	scripts/verify-checksum-models.py	40	# Create an array to store the results	COMMENT

Modern AI Meta-Vocabulary15 hits · 42 pts

Severity	File	Line	Snippet	Context
MEDIUM	…sting _deepseek-ai_DeepSeek-V3-0324_ model support..md	3732	> Oh nice, a single html sounds cool. I want to re-write my little `dchat.py` app to remove litellm dependency and simpl	CODE
MEDIUM	…sting _deepseek-ai_DeepSeek-V3-0324_ model support..md	3776	> >I'd like to add [primp](https://github.com/deedy5/primp) directly to it, which I use for my "agentic" stuff like web	CODE
MEDIUM	…discussions/15 - Will LQER improve k- and i-quants_.md	28	* In `llama.cpp` the mean log probability is evaluated over the second half of the context window `n_ctx`, while in Pyth	COMMENT
MEDIUM	…UDA error_ an illegal memory access was encountered.md	3457	The bug was in the matrix-vector multiplication kernel. It only shows up when the number of rows being processed (i.e.,	CODE
MEDIUM	…server/webui_llamacpp/src/lib/stores/agentic.svelte.ts	395	// Ensure built-in tools are fetched before checking if agentic is enabled	COMMENT
MEDIUM	…es/server/webui_llamacpp/src/lib/stores/chat.svelte.ts	544	// If agentic loop is running, inject as a steering message instead of starting a new flow	COMMENT
MEDIUM	…es/server/webui_llamacpp/src/lib/stores/chat.svelte.ts	550	// If non-agentic streaming is active, queue as a pending message to send after completion	COMMENT
MEDIUM	…es/server/webui_llamacpp/src/lib/stores/chat.svelte.ts	739	// Update the first assistant message with cumulative agentic timings	COMMENT
MEDIUM	…es/server/webui_llamacpp/src/lib/stores/chat.svelte.ts	901	// Generate LLM based title for new conversations after agentic flow completes	COMMENT
MEDIUM	…erver/webui_llamacpp/src/lib/utils/legacy-migration.ts	181	// Has agentic markers - full migration needed	COMMENT
MEDIUM	examples/server/webui_llamacpp/src/lib/utils/agentic.ts	269	// Walk consecutive trailing tool results. The agentic loop only emits tool	COMMENT
MEDIUM	…b/components/app/chat/ChatMessages/ChatMessages.svelte	179	// Build display entries, grouping agentic sessions into single entries.	COMMENT
MEDIUM	…b/components/app/chat/ChatMessages/ChatMessages.svelte	180	// An agentic session = assistant(with tool_calls) → tool → assistant → tool → ... → assistant(final)	COMMENT
MEDIUM	…app/chat/ChatMessages/ChatMessageAgenticContent.svelte	114	// Group flat sections into agentic turns	COMMENT
MEDIUM	examples/gritlm/README.md	7	### Retrieval-Augmented Generation (RAG) use case	COMMENT

Empty Error Handlers7 hits · 35 pts

Severity	File	Line	Snippet	Context
HIGH	convert_imatrix_gguf_to_dat.py	27	except AttributeError:	CODE
HIGH	gguf-py/gguf/vocab.py	366	except FileNotFoundError:	CODE
HIGH	gguf-py/gguf/vocab.py	439	except FileNotFoundError:	CODE
HIGH	gguf-py/gguf/metadata.py	225	except ValueError:	CODE
HIGH	examples/convert_legacy_llama.py	1230	except FileNotFoundError:	CODE
HIGH	scripts/run-with-preset.py	143	except KeyboardInterrupt:	CODE
HIGH	scripts/compare-llama-bench.py	320	except ValueError:	CODE

Redundant / Tautological Comments12 hits · 18 pts

Severity	File	Line	Snippet	Context
LOW	gguf-py/gguf/metadata.py	176	# Check if we erroneously matched against './' or '../' etc...	COMMENT
LOW	gguf-py/gguf/gguf_reader.py	74	# Check if it's unsafe to perform slice optimization on data	COMMENT
LOW	gguf-py/gguf/gguf_reader.py	229	# Check if it's a simple scalar type.	COMMENT
LOW	gguf-py/scripts/gguf_dump.py	334	# Check if new Tensor Group	COMMENT
LOW	examples/convert_legacy_llama.py	1178	# Check if it's a set of safetensors files first	COMMENT
LOW	examples/pydantic_models_to_grammar.py	490	# Check if the field is optional (not required)	COMMENT
LOW	examples/pydantic_models_to_grammar.py	96	if not parts: # Check if the list is empty	CODE
LOW	examples/mtmd/tests.sh	16	# Check if the first argument is "big", then run test with big models	COMMENT
LOW⚡	scripts/verify-checksum-models.py	31	# Check if the hash list file exists	COMMENT
LOW	scripts/verify-checksum-models.py	54	# Check if the file exists	COMMENT
LOW	scripts/verify-checksum-models.py	70	# Add the results to the array	COMMENT
LOW	scripts/verify-checksum-models.py	82	# Output the results as a table	COMMENT

Decorative Section Separators4 hits · 18 pts

Severity	File	Line	Snippet	Context
MEDIUM⚡	examples/mtmd/clip.cpp	6041	// ├─────┤ │	COMMENT
MEDIUM⚡	examples/mtmd/clip.cpp	6043	// ├─────┤ │	COMMENT
MEDIUM⚡	examples/mtmd/clip.cpp	6045	// └─────┘ │	COMMENT
MEDIUM⚡	examples/mtmd/clip.cpp	6046	// ──────┘ x B	COMMENT

Cross-Language Confusion4 hits · 18 pts

Severity	File	Line	Snippet	Context
HIGH	tests/test-tokenizer-random.py	347	if unicodedata.category(chr(cpt)) in ("Cn", "Cs", "Co"): # undefined, surrogates, private	CODE
HIGH	scripts/tool_bench.py	10	cmake -B build -DLLAMA_CURL=1 && cmake --build build --config Release -j -t llama-server	STRING
HIGH	scripts/compare-llama-bench.py	48	$ make clean && make llama-bench	CODE
HIGH	scripts/compare-llama-bench.py	51	$ make clean && make llama-bench	CODE

Example Usage Blocks10 hits · 15 pts

Severity	File	Line	Snippet	Context
LOW	gguf-py/examples/writer.py	13	# Example usage:	COMMENT
LOW	tests/test-tokenizer-0.sh	3	# Usage:	COMMENT
LOW	tests/test-tokenizer-random.py	4	# Sample usage:	COMMENT
LOW	examples/json_schema_pydantic_example.py	1	# Usage:	COMMENT
LOW	examples/base-translate.sh	6	# Usage:	COMMENT
LOW	examples/server/webui_llamacpp/scripts/dev.sh	8	# Usage:	COMMENT
LOW	scripts/sync-ggml-am.sh	5	# Usage:	COMMENT
LOW	scripts/build-zen.sh	6	# Usage:	COMMENT
LOW	scripts/server-llm.sh	16	# Usage:	COMMENT
LOW	scripts/hf.sh	5	# Usage:	COMMENT

AI Response Leakage3 hits · 15 pts

Severity	File	Line	Snippet	Context
HIGH	…pull_requests/565 - add hunyuan moe support for 561.md	177	The mainline PR `https://github.com/ggml-org/llama.cpp/pull/14425` seems still open for now, and yes no rush to merge th	CODE
HIGH	github-data/issues/88 - Bug_ Won_t compile on MSVC.md	15	As mentioned in #82 this does not compile with MSVC. I was able to get through the issues and make it compile on my mach	CODE
HIGH	github-data/issues/345 - build question newbie.md	376	I'm not familiar with this space, so had to look up what "rock 5b" is. According to [this](https://bret.dk/radxa-rock-5b	CODE

Modern Structural Boilerplate12 hits · 12 pts

Severity	File	Line	Snippet	Context
LOW	convert_hf_to_gguf.py	774	def _set_vocab_gpt2(self, dir_model: Path \| None = None, vocab_size: int \| None = None) -> None:	CODE
LOW	convert_hf_to_gguf.py	2482	def _set_vocab_gemma4(self, dir_model: Path, vocab_size: int \| None = None) -> None:	CODE
LOW	convert_hf_to_gguf.py	2732	def _set_vocab_tokenizer_json(self, dir_model: Path, vocab_size: int) -> None:	CODE
LOW	convert_hf_to_gguf.py	2881	def set_gguf_parameters(self) -> None:	CODE
LOW	convert_hf_to_gguf.py	7251	def set_vocab(self) -> None:	CODE
LOW	gguf-py/gguf/gguf_writer.py	34	logger = logging.getLogger(__name__)	CODE
LOW	gguf-py/gguf/vocab.py	39	logger = logging.getLogger(__name__)	CODE
LOW	gguf-py/gguf/vocab.py	129	def _set_special_token(self, typ: str, tid: Any) -> None:	CODE
LOW	gguf-py/gguf/gguf_reader.py	33	logger = logging.getLogger(__name__)	CODE
LOW	gguf-py/gguf/lazy.py	11	logger = logging.getLogger(__name__)	CODE
LOW	gguf-py/scripts/gguf_set_metadata.py	44	def set_metadata(reader: GGUFReader, args: argparse.Namespace) -> None:	CODE
LOW	scripts/tool_bench.py	75	logger = logging.getLogger(__name__)	CODE

AI Slop Vocabulary4 hits · 11 pts

Severity	File	Line	Snippet	Context
MEDIUM	gguf-py/gguf/metadata.py	309	# (Definitely not part of hugging face formal/informal standard)	COMMENT
MEDIUM	gguf-py/gguf/metadata.py	325	# (Definitely not part of hugging face formal/informal standard... but with model_ appended to match their s	COMMENT
MEDIUM	gguf-py/gguf/lazy.py	49	# TODO: make this even more comprehensive	COMMENT
MEDIUM	examples/server/webui_llamacpp/package-lock.json	9611	"resolved": "https://registry.npmjs.org/robust-predicates/-/robust-predicates-3.0.3.tgz",	CODE

Slop Phrases4 hits · 6 pts

Severity	File	Line	Snippet	Context
MEDIUM	scripts/server-llm.sh	194	printf " model using llama.cpp for demonstration purposes.\n"	CODE
MEDIUM	scripts/server-llm.sh	204	printf " - This script is only for demonstration purposes\n"	CODE
LOW	.github/ISSUE_TEMPLATE/07-refactor.yml	9	Don't forget to [check for existing refactor issue tickets](https://github.com/ikawrakow/ik_llama.cpp/issues?q=i	CODE
LOW	.github/ISSUE_TEMPLATE/06-research.yml	9	Don't forget to check for any duplicate research issue tickets	CODE

AI Structural Patterns7 hits · 6 pts

Severity	File	Line	Context
LOW	convert_hf_to_gguf.py	74	CODE
LOW	examples/pydantic_models_to_grammar_examples.py	308	CODE
LOW	examples/pydantic_models_to_grammar.py	1027	CODE
LOW	…egacy-models/minicpmv-convert-image-encoder-to-gguf.py	82	CODE
LOW	examples/server/tests/features/steps/steps.py	1051	CODE
LOW	examples/server/tests/features/steps/steps.py	1097	CODE
LOW	scripts/tool_bench.py	204	CODE

Synthetic Comment Markers1 hit · 5 pts

Severity	File	Line	Snippet	Context
HIGH	…pull_requests/287 - Is this better for DeepSeek-R1_.md	303	The auto-generated code python:	CODE

Docstring Block Structure1 hit · 5 pts

Severity	File	Line	Snippet	Context
HIGH	examples/pydantic_models_to_grammar.py	532	Generate GBNF Grammar from Pydantic Models. This method takes a list of Pydantic models and uses them to gener	STRING

Fake / Example Data4 hits · 4 pts

Severity	File	Line	Snippet	Context
LOW	tests/test-function-calls.cpp	492	const std::string real_world_api_call = R"(I'll make an API call for you. functions.http_request:0{"method": "POST", "ur	CODE
LOW	examples/server/public_mikupad/index.html	3664	// { "placeholder":"replacement" }	COMMENT
LOW	examples/server/public_mikupad/index.html	4162	'placeholder'="Name for saved prompt"	CODE
LOW	examples/server/public_mikupad/index.html	5572	// { "placeholder":"replacement" }	COMMENT

TODO Padding2 hits · 3 pts

Severity	File	Line	Snippet	Context
LOW	common/speculative.cpp	689	// TODO: implement	COMMENT
LOW	examples/server/webui/src/components/ChatMessage.tsx	62	// TODO: implement this as remark/rehype plugin in the future	COMMENT

ikawrakow/ik_llama.cpp

Open-Source Code Quality, Security and Technical Debt Review