ai-dynamo/dynamo

13.5

Adjusted Score

13.5

Raw Score

100%

Time Factor

2026-07-14

Last Push

7.5K

Stars

Rust

Language

1.3M

Lines of Code

4.2K

Files

12.6K

Pattern Hits

2026-07-14

Scan Date

0.03

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 13HIGH 120MEDIUM 1744LOW 10699

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 12576 distinct pattern matches across 26 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 Separators1485 hits · 5302 pts

Severity	File	Line	Snippet	Context
MEDIUM⚡	deploy/power-agent/power_agent.py	114	# ---------------------------------------------------------------------------	COMMENT
MEDIUM⚡	deploy/power-agent/power_agent.py	118	# ---------------------------------------------------------------------------	COMMENT
MEDIUM⚡	deploy/power-agent/power_agent.py	1003	# ---------------------------------------------------------------------------	COMMENT
MEDIUM⚡	deploy/power-agent/power_agent.py	1005	# ---------------------------------------------------------------------------	COMMENT
MEDIUM⚡	deploy/power-agent/power_agent.py	1678	# ---------------------------------------------------------------------------	COMMENT
MEDIUM⚡	deploy/power-agent/power_agent.py	1680	# ---------------------------------------------------------------------------	COMMENT
MEDIUM	deploy/power-agent/power_agent.py	74	# ---------------------------------------------------------------------------	COMMENT
MEDIUM	deploy/power-agent/power_agent.py	76	# ---------------------------------------------------------------------------	COMMENT
MEDIUM	deploy/power-agent/power_agent.py	152	# ---------------------------------------------------------------------------	COMMENT
MEDIUM	deploy/power-agent/power_agent.py	154	# ---------------------------------------------------------------------------	COMMENT
MEDIUM	deploy/power-agent/power_agent.py	366	# ---------------------------------------------------------------------------	COMMENT
MEDIUM	deploy/power-agent/power_agent.py	368	# ---------------------------------------------------------------------------	COMMENT
MEDIUM	deploy/power-agent/power_agent.py	449	# ---------------------------------------------------------------------------	COMMENT
MEDIUM	deploy/power-agent/power_agent.py	451	# ---------------------------------------------------------------------------	COMMENT
MEDIUM	deploy/power-agent/power_agent.py	822	# ---------------------------------------------------------------------------	COMMENT
MEDIUM	deploy/power-agent/power_agent.py	824	# ---------------------------------------------------------------------------	COMMENT
MEDIUM	deploy/power-agent/power_agent.py	850	# ---------------------------------------------------------------------------	COMMENT
MEDIUM	deploy/power-agent/power_agent.py	852	# ---------------------------------------------------------------------------	COMMENT
MEDIUM	deploy/power-agent/power_agent.py	1160	# ---------------------------------------------------------------------------	COMMENT
MEDIUM	deploy/power-agent/power_agent.py	1162	# ---------------------------------------------------------------------------	COMMENT
MEDIUM	deploy/power-agent/power_agent.py	1258	# ---------------------------------------------------------------------------	COMMENT
MEDIUM	deploy/power-agent/power_agent.py	1260	# ---------------------------------------------------------------------------	COMMENT
MEDIUM⚡	deploy/power-agent/actuator.py	764	# ------------------------------------------------------------------	COMMENT
MEDIUM⚡	deploy/power-agent/actuator.py	766	# ------------------------------------------------------------------	COMMENT
MEDIUM⚡	deploy/power-agent/actuator.py	2043	# ------------------------------------------------------------------	COMMENT
MEDIUM⚡	deploy/power-agent/actuator.py	2045	# ------------------------------------------------------------------	COMMENT
MEDIUM	deploy/power-agent/actuator.py	640	# ------------------------------------------------------------------	COMMENT
MEDIUM	deploy/power-agent/actuator.py	642	# ------------------------------------------------------------------	COMMENT
MEDIUM	deploy/power-agent/actuator.py	833	# ------------------------------------------------------------------	COMMENT
MEDIUM	deploy/power-agent/actuator.py	835	# ------------------------------------------------------------------	COMMENT
MEDIUM	deploy/power-agent/actuator.py	1200	# ------------------------------------------------------------------	COMMENT
MEDIUM	deploy/power-agent/actuator.py	1202	# ------------------------------------------------------------------	COMMENT
MEDIUM⚡	deploy/power-agent/tests/test_orphan_recovery.py	101	# ---------------------------------------------------------------------------	COMMENT
MEDIUM⚡	deploy/power-agent/tests/test_orphan_recovery.py	103	# ---------------------------------------------------------------------------	COMMENT
MEDIUM⚡	deploy/power-agent/tests/test_orphan_recovery.py	155	# ---------------------------------------------------------------------------	COMMENT
MEDIUM⚡	deploy/power-agent/tests/test_orphan_recovery.py	157	# ---------------------------------------------------------------------------	COMMENT
MEDIUM⚡	deploy/power-agent/tests/test_orphan_recovery.py	230	# ---------------------------------------------------------------------------	COMMENT
MEDIUM⚡	deploy/power-agent/tests/test_orphan_recovery.py	232	# ---------------------------------------------------------------------------	COMMENT
MEDIUM⚡	deploy/power-agent/tests/test_orphan_recovery.py	269	# ---------------------------------------------------------------------------	COMMENT
MEDIUM⚡	deploy/power-agent/tests/test_orphan_recovery.py	271	# ---------------------------------------------------------------------------	COMMENT
MEDIUM⚡	deploy/power-agent/tests/test_orphan_recovery.py	549	# ---------------------------------------------------------------------------	COMMENT
MEDIUM⚡	deploy/power-agent/tests/test_orphan_recovery.py	551	# ---------------------------------------------------------------------------	COMMENT
MEDIUM⚡	deploy/power-agent/tests/test_orphan_recovery.py	579	# ---------------------------------------------------------------------------	COMMENT
MEDIUM⚡	deploy/power-agent/tests/test_orphan_recovery.py	581	# ---------------------------------------------------------------------------	COMMENT
MEDIUM	deploy/power-agent/tests/test_orphan_recovery.py	41	# ---------------------------------------------------------------------------	COMMENT
MEDIUM	deploy/power-agent/tests/test_orphan_recovery.py	43	# ---------------------------------------------------------------------------	COMMENT
MEDIUM	deploy/power-agent/tests/test_orphan_recovery.py	182	# ---------------------------------------------------------------------------	COMMENT
MEDIUM	deploy/power-agent/tests/test_orphan_recovery.py	184	# ---------------------------------------------------------------------------	COMMENT
MEDIUM	deploy/power-agent/tests/test_orphan_recovery.py	359	# ---------------------------------------------------------------------------	COMMENT
MEDIUM	deploy/power-agent/tests/test_orphan_recovery.py	361	# ---------------------------------------------------------------------------	COMMENT
MEDIUM	deploy/power-agent/tests/test_orphan_recovery.py	419	# ---------------------------------------------------------------------------	COMMENT
MEDIUM	deploy/power-agent/tests/test_orphan_recovery.py	421	# ---------------------------------------------------------------------------	COMMENT
MEDIUM	deploy/power-agent/tests/e2e_actuator_parity.py	62	# ---------------------------------------------------------------------------	COMMENT
MEDIUM	deploy/power-agent/tests/e2e_actuator_parity.py	66	# ---------------------------------------------------------------------------	COMMENT
MEDIUM	deploy/power-agent/tests/e2e_actuator_parity.py	93	# ---------------------------------------------------------------------------	COMMENT
MEDIUM	deploy/power-agent/tests/e2e_actuator_parity.py	97	# ---------------------------------------------------------------------------	COMMENT
MEDIUM	deploy/power-agent/tests/e2e_actuator_parity.py	141	# ---------------------------------------------------------------------------	COMMENT
MEDIUM	deploy/power-agent/tests/e2e_actuator_parity.py	143	# ---------------------------------------------------------------------------	COMMENT
MEDIUM	deploy/power-agent/tests/e2e_actuator_parity.py	223	# ---------------------------------------------------------------------------	COMMENT
MEDIUM	deploy/power-agent/tests/e2e_actuator_parity.py	226	# ---------------------------------------------------------------------------	COMMENT
1425 more matches not shown…

Hyper-Verbose Identifiers4931 hits · 4726 pts

Severity	File	Line	Snippet	Context
LOW	conftest.py	46	def _patched_get_preparation_data(name):	CODE
LOW⚡	deploy/power-agent/power_agent.py	128	def _extract_pod_uid_from_cgroup(pid: int) -> Optional[str]:	CODE
LOW⚡	deploy/power-agent/power_agent.py	1008	def _restore_orphaned_gpus_on_startup(actuator: Actuator) -> None:	CODE
LOW	deploy/power-agent/power_agent.py	250	def _load_previously_managed_gpus() -> set[str]:	CODE
LOW	deploy/power-agent/power_agent.py	297	def _record_managed_gpu_by_uuid(uuid: str) -> None:	CODE
LOW	deploy/power-agent/power_agent.py	337	def _flush_pending_acquisitions() -> None:	CODE
LOW	deploy/power-agent/power_agent.py	519	def _retire_actuator_ownership(actuator: Actuator, uuid: str) -> None:	CODE
LOW	deploy/power-agent/power_agent.py	580	def _flush_pending_retirements() -> None:	CODE
LOW	deploy/power-agent/actuator.py	1170	def _coerce_power_limit_watts(value: Any, field_name: str, gpu_idx: int) -> int:	CODE
LOW	deploy/power-agent/tests/test_multi_pod_policy.py	59	def safe_default_applied_total(self):	CODE
LOW	deploy/power-agent/tests/test_multi_pod_policy.py	77	def test_single_pod_apply_annotation(self):	CODE
LOW	deploy/power-agent/tests/test_multi_pod_policy.py	101	def test_no_parseable_annotation(self):	CODE
LOW⚡	deploy/power-agent/tests/test_multi_pod_policy.py	113	def test_invalid_annotation_value(self):	CODE
LOW⚡	deploy/power-agent/tests/test_multi_pod_policy.py	121	def test_three_pods_all_agree(self):	CODE
LOW⚡	deploy/power-agent/tests/test_multi_pod_policy.py	131	def test_multipod_missing_annotation_on_one_pod_is_conflict(self):	CODE
LOW	deploy/power-agent/tests/test_multi_pod_policy.py	153	def test_multipod_invalid_annotation_on_one_pod_is_conflict(self):	CODE
LOW	deploy/power-agent/tests/test_multi_pod_policy.py	167	def test_multipod_all_missing_is_conflict(self):	CODE
LOW⚡	deploy/power-agent/tests/test_orphan_recovery.py	107	def test_unmanaged_uuid_is_skipped(self):	CODE
LOW⚡	deploy/power-agent/tests/test_orphan_recovery.py	161	def test_gpu_with_running_pids_is_skipped(self):	CODE
LOW⚡	deploy/power-agent/tests/test_orphan_recovery.py	236	def test_one_gpu_failure_does_not_abort_others(self):	CODE
LOW⚡	deploy/power-agent/tests/test_orphan_recovery.py	275	def test_successful_restore_discards_uuid(self):	CODE
LOW⚡	deploy/power-agent/tests/test_orphan_recovery.py	560	def test_inconclusive_read_skips_recovery_and_does_not_rewrite(self):	CODE
LOW⚡	deploy/power-agent/tests/test_orphan_recovery.py	590	def test_persist_failure_queues_retirements_and_does_not_raise(self):	CODE
LOW	deploy/power-agent/tests/test_orphan_recovery.py	140	def test_no_managed_uuids_means_no_writes(self):	CODE
LOW	deploy/power-agent/tests/test_orphan_recovery.py	195	def test_idle_managed_gpu_delegates_to_restore_by_uuid(self):	CODE
LOW	deploy/power-agent/tests/test_orphan_recovery.py	211	def test_caller_does_not_read_current_or_default(self):	CODE
LOW	deploy/power-agent/tests/test_orphan_recovery.py	291	def test_at_default_gpu_retires_ownership(self):	CODE
LOW	deploy/power-agent/tests/test_orphan_recovery.py	319	def test_retired_uuid_is_not_clobbered_on_next_startup(self):	CODE
LOW	deploy/power-agent/tests/test_orphan_recovery.py	381	def test_true_return_discards_uuid(self):	CODE
LOW	deploy/power-agent/tests/test_orphan_recovery.py	391	def test_none_return_retires_ownership(self):	CODE
LOW	deploy/power-agent/tests/test_orphan_recovery.py	405	def test_false_return_retains_uuid(self):	CODE
LOW	deploy/power-agent/tests/test_orphan_recovery.py	436	def test_absent_uuid_is_pruned_on_complete_scan_without_hardware_write(self):	CODE
LOW	deploy/power-agent/tests/test_orphan_recovery.py	462	def test_absent_uuid_pruned_when_no_devices_present(self):	CODE
LOW	deploy/power-agent/tests/test_orphan_recovery.py	474	def test_incomplete_scan_retains_unobserved_uuid_and_skips_hardware(self):	CODE
LOW	deploy/power-agent/tests/test_orphan_recovery.py	495	def test_busy_gpu_behind_inconclusive_scan_keeps_its_cap(self):	CODE
LOW	deploy/power-agent/tests/test_orphan_recovery.py	521	def test_uuid_moved_to_new_index_after_growth_is_not_pruned(self):	CODE
LOW	deploy/power-agent/tests/test_orphan_recovery.py	609	def test_reconcile_flush_persists_queued_startup_retirement(self):	CODE
LOW	deploy/power-agent/tests/test_shutdown.py	34	def test_fresh_module_copy_shares_managed_state(self):	CODE
LOW	deploy/power-agent/tests/test_shutdown.py	71	def test_handler_sets_shutdown_event(self):	CODE
LOW	deploy/power-agent/tests/test_shutdown.py	76	def test_handler_does_no_cleanup_work(self):	CODE
LOW	deploy/power-agent/tests/test_shutdown.py	96	def test_cleanup_runs_once_on_normal_shutdown(self):	CODE
LOW	deploy/power-agent/tests/test_shutdown.py	108	def test_cleanup_still_runs_when_reconcile_raises(self):	CODE
LOW	deploy/power-agent/tests/test_shutdown.py	128	def test_restores_default_via_actuator_for_each_managed_gpu(self):	CODE
LOW	deploy/power-agent/tests/test_shutdown.py	141	def test_dcgm_actuator_routes_restore_through_dcgm(self):	CODE
LOW	deploy/power-agent/tests/test_shutdown.py	154	def test_per_gpu_restore_failure_does_not_prevent_shutdown(self):	CODE
LOW	deploy/power-agent/tests/test_shutdown.py	167	def test_actuator_shutdown_failure_is_logged_not_raised(self):	CODE
LOW	deploy/power-agent/tests/test_shutdown.py	181	def test_no_managed_gpus_still_shuts_actuator_down(self):	CODE
LOW	deploy/power-agent/tests/test_shutdown.py	210	def test_successful_restore_prunes_uuid_and_persists(self):	CODE
LOW	deploy/power-agent/tests/test_shutdown.py	226	def test_successful_restore_prunes_actuator_managed_uuid_when_available(self):	CODE
LOW	deploy/power-agent/tests/test_shutdown.py	253	def test_skipped_restore_does_not_prune(self):	CODE
LOW	deploy/power-agent/tests/test_shutdown.py	270	def test_failed_restore_does_not_prune(self):	CODE
LOW	deploy/power-agent/tests/test_shutdown.py	291	def test_uuid_lookup_failure_is_logged_but_not_fatal(self):	CODE
LOW	deploy/power-agent/tests/test_shutdown.py	309	def test_persist_failure_does_not_prevent_shutdown(self):	CODE
LOW	deploy/power-agent/tests/test_shutdown.py	382	def test_sweep_restores_leaked_uuid_not_in_index_set(self):	CODE
LOW	deploy/power-agent/tests/test_shutdown.py	399	def test_sweep_ignores_unowned_persisted_uuid(self):	CODE
LOW	deploy/power-agent/tests/test_shutdown.py	416	def test_sweep_retires_owned_uuid_already_at_default(self):	CODE
LOW	deploy/power-agent/tests/test_shutdown.py	430	def test_sweep_keeps_uuid_on_inconclusive_scan(self):	CODE
LOW	deploy/power-agent/tests/test_shutdown.py	444	def test_sweep_exception_keeps_uuid_and_still_shuts_down(self):	CODE
LOW	deploy/power-agent/tests/test_shutdown.py	458	def test_indexed_restore_retires_uuid_before_sweep(self):	CODE
LOW	deploy/power-agent/tests/test_shutdown.py	507	def test_sweep_does_not_run_on_nvml_actuator(self):	CODE
4871 more matches not shown…

Over-Commented Block2222 hits · 2155 pts

Severity	File	Line	Snippet	Context
LOW	deny.toml	1	# SPDX-FileCopyrightText: Copyright (c) 2024-2026 NVIDIA CORPORATION & AFFILIATES. All rights reserved.	COMMENT
LOW	.pre-commit-config.yaml	1	# SPDX-FileCopyrightText: Copyright (c) 2024-2026 NVIDIA CORPORATION & AFFILIATES. All rights reserved.	COMMENT
LOW	pyproject.toml	201	# chardet 6.x is pulled in transitively but requests 2.32.x only supports chardet <6;	COMMENT
LOW	deploy/__init__.py	1	# SPDX-FileCopyrightText: Copyright (c) 2025-2026 NVIDIA CORPORATION & AFFILIATES. All rights reserved.	COMMENT
LOW	…ploy/snapshot/plugins/inet-remap/snapshot_inet_remap.c	1	// SPDX-FileCopyrightText: Copyright (c) 2026 NVIDIA CORPORATION & AFFILIATES. All rights reserved.	COMMENT
LOW	…loy/snapshot/internal/controller/podsnapshotcontent.go	41	// CheckpointID is the stable artifact identity.	COMMENT
LOW	deploy/snapshot/protocol/control_volume.go	1	// SPDX-FileCopyrightText: Copyright (c) 2025-2026 NVIDIA CORPORATION & AFFILIATES. All rights reserved.	COMMENT
LOW	deploy/snapshot/protocol/control_volume.go	21		COMMENT
LOW	deploy/snapshot/protocol/control_volume.go	41	// checkpoint. Observed by the checkpoint job's kubelet readiness probe	COMMENT
LOW	deploy/power-agent/pytest.ini	1	# Standalone pytest config for the power-agent deployable unit.	COMMENT
LOW	deploy/power-agent/managed_state.py	61	# currently OWNS a below-default cap on — added on a successful cap write and	COMMENT
LOW	deploy/power-agent/power_agent.py	81	# Pod-LIST timeouts. These exist so a SIGTERM that lands while the reconcile	COMMENT
LOW	deploy/power-agent/power_agent.py	101	# exact wall-clock bound. This is an always-on steady-state timeout, not a	COMMENT
LOW	deploy/power-agent/power_agent.py	281		COMMENT
LOW	deploy/power-agent/power_agent.py	701	# membership alone — restoring the wrong GPU and discarding the real	COMMENT
LOW	deploy/power-agent/power_agent.py	1061	# per-index `get_uuid` loop. That naive loop was falsely conclusive on the	COMMENT
LOW	deploy/power-agent/power_agent.py	1081	# re-enumerated GPU's workload and let us restore/retire the wrong	COMMENT
LOW	deploy/power-agent/power_agent.py	1281	raise RuntimeError("kubernetes Python SDK is required — install kubernetes")	COMMENT
LOW	deploy/power-agent/power_agent.py	1341	# apiserver request per node per cycle, so aggregate request rate	COMMENT
LOW	deploy/power-agent/power_agent.py	1441	return	COMMENT
LOW	deploy/power-agent/power_agent.py	1541	(the opted-in pod is gone and a non-managed tenant owns the GPU,	COMMENT
LOW	deploy/power-agent/power_agent.py	1621	# No opted-in pod owns this GPU (every process is either non-K8s or	COMMENT
LOW	deploy/power-agent/power_agent.py	1781	# while the actuator reaches the agent via `import power_agent` — two	COMMENT
LOW	deploy/power-agent/actuator.py	601	# init() would leak one refcount per reconnect and leave NVML	COMMENT
LOW	deploy/power-agent/actuator.py	621	# In-process restore identity snapshot. `_managed_gpu_indices` stores	COMMENT
LOW	deploy/power-agent/actuator.py	821		COMMENT
LOW	deploy/power-agent/actuator.py	1241	# establish which physical GPU that is, we cannot guarantee the cap	COMMENT
LOW	deploy/power-agent/actuator.py	1261	gpu_idx,	COMMENT
LOW	deploy/power-agent/actuator.py	1421	# the Set, and no post-Set reconnect-capable re-read that a	COMMENT
LOW	deploy/power-agent/actuator.py	1501		COMMENT
LOW	deploy/power-agent/actuator.py	1581	# Guard the write with the UUID we intend to restore so a reconnect	COMMENT
LOW	deploy/power-agent/tests/test_dcgm_actuator.py	901	nvml = MagicMock()	COMMENT
LOW	deploy/utils/dynamo_deployment.py	1	# SPDX-FileCopyrightText: Copyright (c) 2025-2026 NVIDIA CORPORATION & AFFILIATES. All rights reserved.	COMMENT
LOW	deploy/utils/__init__.py	1	# SPDX-FileCopyrightText: Copyright (c) 2025-2026 NVIDIA CORPORATION & AFFILIATES. All rights reserved.	COMMENT
LOW	deploy/utils/kubernetes.py	1	# SPDX-FileCopyrightText: Copyright (c) 2025-2026 NVIDIA CORPORATION & AFFILIATES. All rights reserved.	COMMENT
LOW	deploy/utils/convert_api_version.py	1	# SPDX-FileCopyrightText: Copyright (c) 2025-2026 NVIDIA CORPORATION & AFFILIATES. All rights reserved.	COMMENT
LOW	deploy/utils/tests/__init__.py	1	# SPDX-FileCopyrightText: Copyright (c) 2026 NVIDIA CORPORATION & AFFILIATES. All rights reserved.	COMMENT
LOW	deploy/utils/tests/test_convert_api_version.py	1	# SPDX-FileCopyrightText: Copyright (c) 2025-2026 NVIDIA CORPORATION & AFFILIATES. All rights reserved.	COMMENT
LOW	deploy/pre-deployment/pre-deployment-check.sh	1	#!/usr/bin/env bash	COMMENT
LOW	deploy/inference-gateway/epp/cmd/epp/main.go	21	//	COMMENT
LOW	…ploy/inference-gateway/ext-proc/tests/ext_proc_test.rs	1	// SPDX-FileCopyrightText: Copyright (c) 2024-2026 NVIDIA CORPORATION & AFFILIATES. All rights reserved.	COMMENT
LOW	deploy/inference-gateway/ext-proc/src/envoy_helpers.rs	41	"content-length",	COMMENT
LOW	deploy/inference-gateway/ext-proc/src/epp.rs	181	// unused on this path (matches the prior namespace-lookup miss).	COMMENT
LOW	deploy/inference-gateway/ext-proc/src/epp.rs	641	actual_namespace,	COMMENT
LOW	deploy/inference-gateway/ext-proc/src/lib.rs	1	// SPDX-FileCopyrightText: Copyright (c) 2024-2026 NVIDIA CORPORATION & AFFILIATES. All rights reserved.	COMMENT
LOW	deploy/inference-gateway/ext-proc/src/server.rs	561	});	COMMENT
LOW	deploy/inference-gateway/ext-proc/src/server.rs	581	///	COMMENT
LOW	deploy/inference-gateway/ext-proc/src/picker.rs	1	// SPDX-FileCopyrightText: Copyright (c) 2024-2026 NVIDIA CORPORATION & AFFILIATES. All rights reserved.	COMMENT
LOW	deploy/inference-gateway/ext-proc/src/picker.rs	41	/// header keys (valid in HTTP) are not silently collapsed.	COMMENT
LOW	deploy/inference-gateway/ext-proc/src/picker.rs	61	/// Used by Dynamo for routing metadata (worker IDs, DP ranks, routing mode).	COMMENT
LOW	…rence-gateway/scripts/install_gaie_crd_agentgateway.sh	1	#!/usr/bin/env bash	COMMENT
LOW	…rence-gateway/scripts/install_gaie_crd_agentgateway.sh	41	helm upgrade -i --create-namespace --namespace "$AGW_NAMESPACE" --version "$AGW_VERSION" \	COMMENT
LOW	deploy/helm/charts/snapshot/Chart.yaml	1	# SPDX-FileCopyrightText: Copyright (c) 2025-2026 NVIDIA CORPORATION & AFFILIATES. All rights reserved.	COMMENT
LOW	deploy/helm/charts/snapshot/values.yaml	1	# SPDX-FileCopyrightText: Copyright (c) 2025-2026 NVIDIA CORPORATION & AFFILIATES. All rights reserved.	COMMENT
LOW	deploy/helm/charts/snapshot/values.yaml	41	# requires ReadWriteMany storage because DaemonSet pods on multiple nodes	COMMENT
LOW	deploy/helm/charts/snapshot/values.yaml	141	# Create service account	COMMENT
LOW	deploy/helm/charts/power-agent/Chart.yaml	21		COMMENT
LOW	deploy/helm/charts/power-agent/values.yaml	1	# SPDX-FileCopyrightText: Copyright (c) 2025-2026 NVIDIA CORPORATION & AFFILIATES. All rights reserved.	COMMENT
LOW	deploy/helm/charts/power-agent/values.yaml	21	# be re-pointed by the registry. Must look like `sha256:<hex>`.	COMMENT
LOW	deploy/helm/charts/power-agent/values.yaml	41	# default (not this floor); otherwise the GPU is left untouched.	COMMENT
2162 more matches not shown…

Excessive Try-Catch Wrapping866 hits · 850 pts

Severity	File	Line	Snippet	Context
LOW	conftest.py	51	except Exception:	CODE
LOW⚡	deploy/power-agent/power_agent.py	987	except Exception as e:	CODE
LOW⚡	deploy/power-agent/power_agent.py	995	except Exception:	CODE
LOW⚡	deploy/power-agent/power_agent.py	1669	except Exception as e:	CODE
LOW	deploy/power-agent/power_agent.py	320	except Exception as e:	CODE
LOW	deploy/power-agent/power_agent.py	351	except Exception as e:	CODE
LOW	deploy/power-agent/power_agent.py	437	except Exception as e:	CODE
LOW	deploy/power-agent/power_agent.py	569	except Exception as e:	CODE
LOW	deploy/power-agent/power_agent.py	593	except Exception as e:	CODE
LOW	deploy/power-agent/power_agent.py	660	except Exception as e:	CODE
LOW	deploy/power-agent/power_agent.py	696	except Exception as e:	CODE
LOW	deploy/power-agent/power_agent.py	787	except Exception as e:	CODE
LOW	deploy/power-agent/power_agent.py	917	except Exception as e:	CODE
LOW	deploy/power-agent/power_agent.py	926	except Exception as e:	CODE
LOW	deploy/power-agent/power_agent.py	942	except Exception as e:	CODE
LOW	deploy/power-agent/power_agent.py	966	except Exception as e:	CODE
LOW	deploy/power-agent/power_agent.py	1118	except Exception as e:	CODE
LOW	deploy/power-agent/power_agent.py	1148	except Exception as e:	CODE
LOW	deploy/power-agent/power_agent.py	1385	except Exception as e:	CODE
LOW	deploy/power-agent/power_agent.py	1479	except Exception as e:	CODE
LOW	deploy/power-agent/power_agent.py	1502	except Exception as e:	CODE
LOW	deploy/power-agent/power_agent.py	1563	except Exception as e:	CODE
LOW⚡	deploy/power-agent/actuator.py	756	except Exception:	CODE
LOW⚡	deploy/power-agent/actuator.py	2034	except Exception as e:	CODE
LOW	deploy/power-agent/actuator.py	302	except Exception:	CODE
LOW	deploy/power-agent/actuator.py	329	except Exception as e:	CODE
LOW	deploy/power-agent/actuator.py	342	except Exception as e:	CODE
LOW	deploy/power-agent/actuator.py	489	except Exception as e:	CODE
LOW	deploy/power-agent/actuator.py	726	except Exception:	CODE
LOW	deploy/power-agent/actuator.py	735	except Exception:	CODE
LOW	deploy/power-agent/actuator.py	806	except Exception:	CODE
LOW	deploy/power-agent/actuator.py	871	except Exception as e:	CODE
LOW	deploy/power-agent/actuator.py	1254	except Exception as e:	CODE
LOW	deploy/power-agent/actuator.py	1438	except Exception as e:	CODE
LOW	deploy/power-agent/actuator.py	1560	except Exception as e:	CODE
LOW	deploy/power-agent/actuator.py	1739	except Exception as e:	CODE
LOW	deploy/power-agent/actuator.py	1851	except Exception as e:	CODE
LOW	deploy/power-agent/actuator.py	1954	except Exception as e:	CODE
LOW	deploy/power-agent/actuator.py	2022	except Exception as e:	CODE
LOW	deploy/power-agent/tests/e2e_actuator_parity.py	480	except Exception as e:	CODE
LOW	deploy/power-agent/tests/e2e_actuator_parity.py	509	except Exception:	CODE
LOW	deploy/utils/dynamo_deployment.py	156	except Exception:	CODE
LOW	deploy/utils/dynamo_deployment.py	336	except Exception:	CODE
LOW	deploy/utils/dynamo_deployment.py	512	except Exception as e:	CODE
MEDIUM	deploy/utils/dynamo_deployment.py	616	print(f"Error getting logs for pod {pod.metadata.name}: {e}")	CODE
LOW	deploy/utils/dynamo_deployment.py	659	except Exception as e:	CODE
LOW	deploy/utils/kubernetes.py	98	except Exception:	CODE
LOW	deploy/utils/kubernetes.py	103	except Exception as e:	CODE
LOW	deploy/utils/kubernetes.py	134	except Exception:	CODE
LOW	deploy/utils/kubernetes.py	184	except Exception as e:	CODE
LOW	deploy/operator/internal/dynamo/backend_vllm.go	153	except Exception as e:	CODE
LOW	deploy/operator/internal/dynamo/backend_vllm.go	171	except Exception as e:	CODE
MEDIUM	deploy/operator/internal/dynamo/backend_vllm.go	128	def leader_pod_is_healthy():	CODE
MEDIUM	…ploy/operator/api/scripts/generate_pydantic_from_go.py	643	print(f"Error: Input file not found: {args.input}")	CODE
MEDIUM	tests/conftest.py	876	def check():	CODE
LOW	tests/conftest.py	493	except Exception as exc:	CODE
LOW	tests/conftest.py	637	except Exception as exc:	CODE
LOW	tests/conftest.py	831	except Exception as e:	CODE
LOW	tests/conftest.py	890	except Exception:	CODE
LOW	tests/conftest.py	911	except Exception as e:	CODE
806 more matches not shown…

Structural Annotation Overuse473 hits · 829 pts

Severity	File	Line	Snippet	Context
LOW⚡	deploy/pre-deployment/nixl/README.md	97	### Step 1: Build nixlbench Docker Image	COMMENT
LOW⚡	deploy/pre-deployment/nixl/README.md	104	### Step 2: Update Deployment YAML File	COMMENT
LOW⚡	deploy/pre-deployment/nixl/README.md	110	### Step 3: Deploy to Kubernetes	COMMENT
LOW⚡	deploy/observability/cleanup-monitoring.sh	18	# Step 1: Delete Grafana dashboard ConfigMap	COMMENT
LOW⚡	deploy/observability/cleanup-monitoring.sh	23	# Step 2: Revert DCGM custom metrics configuration	COMMENT
LOW	deploy/observability/cleanup-monitoring.sh	43	# Step 3: Revert Dynamo operator Prometheus endpoint	COMMENT
LOW⚡	deploy/observability/cleanup-monitoring.sh	64	# Step 4: Uninstall kube-prometheus-stack	COMMENT
LOW⚡	deploy/observability/cleanup-monitoring.sh	72	# Step 5: Delete monitoring namespace (optional)	COMMENT
LOW⚡	deploy/observability/setup-monitoring.sh	20	# Step 1: Add Helm repositories	COMMENT
LOW⚡	deploy/observability/setup-monitoring.sh	27	# Step 2: Update Helm repositories	COMMENT
LOW⚡	deploy/observability/setup-monitoring.sh	32	# Step 3: Install kube-prometheus-stack	COMMENT
LOW⚡	deploy/observability/setup-monitoring.sh	67	# Step 4: Wait for pods to be ready	COMMENT
LOW⚡	deploy/observability/setup-monitoring.sh	73	# Step 5: Verify installation	COMMENT
LOW⚡	deploy/observability/setup-monitoring.sh	78	# Step 6: Update Dynamo operator with Prometheus endpoint	COMMENT
LOW	deploy/observability/setup-monitoring.sh	102	# Step 7: Configure DCGM custom metrics for NVLink profiling	COMMENT
LOW	deploy/observability/setup-monitoring.sh	123	# Step 8: Verify DCGM ServiceMonitor exists (created by GPU operator during cluster setup)	COMMENT
LOW⚡	deploy/observability/setup-monitoring.sh	136	# Step 9: Deploy Grafana dashboard ConfigMap	COMMENT
LOW⚡	deploy/observability/setup-monitoring.sh	142	# Step 10: Get Grafana credentials	COMMENT
LOW	deploy/operator/test/e2e/dgdr/dgdr_test.go	257	// Step 1: Profile without deploying	COMMENT
LOW	deploy/operator/test/e2e/dgdr/dgdr_test.go	276	// Step 2: Verify the output ConfigMap is accessible after profiling	COMMENT
LOW⚡	…loy/operator/test/e2e/dgdr/dgdr_lifecycle_spec_test.go	107	// Step 1: Create DGDR	COMMENT
LOW⚡	…loy/operator/test/e2e/dgdr/dgdr_lifecycle_spec_test.go	115	// Step 2: Wait for profiling to complete (DGDR reaches Ready or later)	COMMENT
LOW⚡	…loy/operator/test/e2e/dgdr/dgdr_lifecycle_spec_test.go	125	// Step 3: Verify ConfigMap output	COMMENT
LOW	…loy/operator/test/e2e/dgdr/dgdr_lifecycle_spec_test.go	147	// Step 4: If autoApply, wait for deployment	COMMENT
LOW	…loy/operator/test/e2e/dgdr/dgdr_lifecycle_spec_test.go	161	// Step 5: Verify DGD reaches successful state	COMMENT
LOW⚡	…loy/operator/test/e2e/dgdr/dgdr_lifecycle_spec_test.go	176	// Step 6: Verify actual pod readiness (independent of DGD status)	COMMENT
LOW⚡	…loy/operator/test/e2e/dgdr/dgdr_lifecycle_spec_test.go	180	// Step 7: Verify individual DGD services	COMMENT
LOW⚡	…loy/operator/test/e2e/dgdr/dgdr_lifecycle_spec_test.go	186	// Step 8: Inference smoke test	COMMENT
LOW	deploy/operator/internal/dynamo/backend_vllm.go	316	// Step 1: Check explicit override	COMMENT
LOW	deploy/operator/internal/dynamo/backend_vllm.go	331	// Step 2: Check operator origin version gate	COMMENT
LOW	…ontroller/dynamocomponentdeployment_controller_test.go	1600	// Step 1: Create the deployment with 1 replica	COMMENT
LOW	…ontroller/dynamocomponentdeployment_controller_test.go	1614	// Step 2: Manually update the deployment to 2 replicas (simulating manual edit)	COMMENT
LOW⚡	…ontroller/dynamocomponentdeployment_controller_test.go	1632	// Step 3: Call createOrUpdateOrDeleteDeployments again - it should reconcile back to 1 replica	COMMENT
LOW⚡	…ontroller/dynamocomponentdeployment_controller_test.go	1638	// Step 4: Verify the deployment was reconciled back to 1 replica	COMMENT
LOW⚡	…ontroller/dynamocomponentdeployment_controller_test.go	1645	// Step 5: Call createOrUpdateOrDeleteDeployments again - it should not be modified	COMMENT
LOW	…/api/v1alpha1/dynamographdeployment_conversion_test.go	1887	// Step 1: first apply. Server ConvertsFrom the user's v1beta1 to	COMMENT
LOW	…/api/v1alpha1/dynamographdeployment_conversion_test.go	1902	// Step 2: second apply. Server reads stored, converts to v1beta1 so	COMMENT
LOW⚡	…/api/v1alpha1/dynamographdeployment_conversion_test.go	1913	// Step 3: apply JSON merge patch. Arrays are replaced wholesale.	COMMENT
LOW⚡	…/api/v1alpha1/dynamographdeployment_conversion_test.go	1923	// Step 4: server converts back to v1alpha1 for storage. This is	COMMENT
LOW⚡	…/api/v1alpha1/dynamographdeployment_conversion_test.go	1930	// Step 5: apiserver generation-bump check.	COMMENT
LOW	…/api/v1alpha1/dynamographdeployment_conversion_test.go	1948	// Step 6: mimic the apiserver's unstructured-level comparison. The	COMMENT
LOW⚡	…alpha1/dynamographdeploymentrequest_conversion_test.go	260	// Step 1: v1alpha1 → v1beta1	COMMENT
LOW⚡	…alpha1/dynamographdeploymentrequest_conversion_test.go	266	// Step 2: v1beta1 → v1alpha1	COMMENT
LOW⚡	…alpha1/dynamographdeploymentrequest_conversion_test.go	319	// Step 1: v1beta1 → v1alpha1	COMMENT
LOW⚡	…alpha1/dynamographdeploymentrequest_conversion_test.go	325	// Step 2: v1alpha1 → v1beta1	COMMENT
LOW⚡	…ploy/operator/api/scripts/generate_pydantic_from_go.py	302	# Step 1: find the index just after the last non-comment line before the block	COMMENT
LOW⚡	…ploy/operator/api/scripts/generate_pydantic_from_go.py	310	# Step 2: forward scan — collect the first paragraph, skip markers	COMMENT
LOW⚡	tests/utils/managed_process.py	571	# Step 3: poll for all process groups to exit.	COMMENT
LOW	tests/utils/managed_process.py	532	# Step 1: Snapshot all process groups before signaling. Only self.proc runs	COMMENT
LOW	tests/utils/managed_process.py	552	# Step 2: SIGTERM all snapshotted process groups (graceful shutdown).	COMMENT
LOW	tests/utils/managed_process.py	594	# Step 4: SIGKILL all snapshotted process groups to ensure nothing	COMMENT
LOW	tests/utils/managed_process.py	831	# WARNING: This method is NOT pytest-xdist safe! Kills ALL matching processes system-wide.	COMMENT
LOW⚡	tests/fault_tolerance/test_vllm_health_check.py	151	# Step 1: Start the frontend	COMMENT
LOW⚡	tests/fault_tolerance/test_vllm_health_check.py	156	# Step 2: Start a worker	COMMENT
LOW⚡	tests/fault_tolerance/test_vllm_health_check.py	163	# Step 3: Send a test request to prove the worker is live.	COMMENT
LOW⚡	tests/fault_tolerance/test_vllm_health_check.py	168	# Step 4: Find and kill vLLM engine processes to force the EngineDeadError condition.	COMMENT
LOW⚡	tests/fault_tolerance/test_vllm_health_check.py	182	# Step 5: Send a request triggering the handler to shutdown everything.	COMMENT
LOW⚡	tests/fault_tolerance/test_vllm_health_check.py	186	# Step 6: Ensure the worker process has been stopped as a result of the EngineDeadError condition.	COMMENT
LOW⚡	tests/fault_tolerance/test_vllm_health_check.py	210	# Step 1: Start the frontend	COMMENT
LOW⚡	tests/fault_tolerance/test_vllm_health_check.py	215	# Step 2: Start a worker	COMMENT
413 more matches not shown…

Unused Imports788 hits · 748 pts

Severity	File	Line	Context
LOW	conftest.py	11	CODE
LOW	deploy/power-agent/actuator.py	39	CODE
LOW	deploy/power-agent/tests/test_orphan_recovery.py	33	CODE
LOW	deploy/power-agent/tests/e2e_actuator_parity.py	43	CODE
LOW	deploy/power-agent/tests/test_managed_state_parser.py	35	CODE
LOW	deploy/power-agent/tests/test_reconcile_wiring.py	27	CODE
LOW	deploy/power-agent/tests/test_multi_dgd_topology.py	37	CODE
LOW	deploy/power-agent/tests/test_dcgm_actuator.py	34	CODE
LOW	tests/report_pytest_markers.py	14	CODE
LOW	tests/frontend/realtime_omni_mock_worker.py	14	CODE
LOW	tests/frontend/test_vllm_prepost_integration.py	4	CODE
LOW	tests/frontend/test_completion_mocker_engine.py	11	CODE
LOW	tests/frontend/test_vllm.py	12	CODE
LOW	tests/frontend/test_request_tracing_logs.py	14	CODE
LOW	tests/frontend/realtime_echo_worker.py	13	CODE
LOW	tests/frontend/test_realtime_python_bridge.py	18	CODE
LOW	tests/frontend/test_unified_worker_otlp_export.py	14	CODE
LOW	tests/frontend/vllm_prepost_worker.py	6	CODE
LOW	tests/frontend/test_realtime_omni_bridge.py	19	CODE
LOW	tests/frontend/test_self_host_metadata.py	17	CODE
LOW	tests/frontend/http_status_propagation_worker.py	7	CODE
LOW	tests/frontend/test_prompt_embeds.py	20	CODE
LOW	tests/frontend/test_http_status_propagation.py	12	CODE
LOW	tests/frontend/test_unified_worker_tracing_smoke.py	17	CODE
LOW	tests/frontend/test_tool_calling_sglang.py	16	CODE
LOW	tests/frontend/grpc/test_tensor_mocker_engine.py	13	CODE
LOW	tests/gpu_memory_service/conftest.py	11	CODE
LOW	tests/gpu_memory_service/test_shadow_failover.py	4	CODE
LOW	tests/gpu_memory_service/flow_assertions.py	4	CODE
LOW	tests/gpu_memory_service/test_pause_resume.py	4	CODE
LOW	tests/gpu_memory_service/common/runtime.py	6	CODE
LOW	tests/gpu_memory_service/common/_gms_server_runner.py	12	CODE
LOW	tests/gpu_memory_service/common/gms.py	6	CODE
LOW	tests/runtime/test_sample_multimodal_smoke.py	6	CODE
LOW	tests/utils/gpu_args.py	6	CODE
LOW	tests/utils/vram_utils.py	21	CODE
LOW	tests/utils/collection_env_guard.py	6	CODE
LOW	tests/utils/pytest_parallel_gpu.py	26	CODE
LOW	tests/utils/router_logs.py	6	CODE
LOW	tests/utils/test_pytest_parallel_gpu.py	12	CODE
LOW	tests/wheels/smoke_install.py	4	CODE
LOW	tests/wheels/test_dynamo_wheel_smoke.py	17	CODE
LOW	tests/fault_tolerance/test_unified_canary.py	6	CODE
LOW	tests/fault_tolerance/test_canary_rank_pause.py	11	CODE
LOW	…e/hardware/fault_injection_service/helpers/__init__.py	26	CODE
LOW	…e/hardware/fault_injection_service/helpers/__init__.py	26	CODE
LOW	…e/hardware/fault_injection_service/helpers/__init__.py	26	CODE
LOW	…e/hardware/fault_injection_service/helpers/__init__.py	26	CODE
LOW	…e/hardware/fault_injection_service/helpers/__init__.py	26	CODE
LOW	…e/hardware/fault_injection_service/helpers/__init__.py	33	CODE
LOW	…e/hardware/fault_injection_service/helpers/__init__.py	33	CODE
LOW	…e/hardware/fault_injection_service/helpers/__init__.py	34	CODE
LOW	…e/hardware/fault_injection_service/helpers/__init__.py	34	CODE
LOW	tests/fault_tolerance/deploy/scenarios.py	30	CODE
LOW	tests/benchmarks/multimodal/jsonl/test_main.py	4	CODE
LOW	tests/benchmarks/multimodal/sweep/test_orchestrator.py	4	CODE
LOW	tests/vllm_self_benchmark/test_self_benchmark_gpu.py	56	CODE
LOW	tests/serve/test_sglang_mm_hashes_protocol.py	25	CODE
LOW	tests/serve/test_vllm_omni.py	12	CODE
LOW	tests/serve/lora_utils.py	29	CODE
728 more matches not shown…

Verbosity Indicators343 hits · 630 pts

Severity	File	Line	Snippet	Context
LOW⚡	deploy/observability/cleanup-monitoring.sh	18	# Step 1: Delete Grafana dashboard ConfigMap	COMMENT
LOW⚡	deploy/observability/cleanup-monitoring.sh	23	# Step 2: Revert DCGM custom metrics configuration	COMMENT
LOW	deploy/observability/cleanup-monitoring.sh	43	# Step 3: Revert Dynamo operator Prometheus endpoint	COMMENT
LOW⚡	deploy/observability/cleanup-monitoring.sh	64	# Step 4: Uninstall kube-prometheus-stack	COMMENT
LOW⚡	deploy/observability/cleanup-monitoring.sh	72	# Step 5: Delete monitoring namespace (optional)	COMMENT
LOW⚡	deploy/observability/setup-monitoring.sh	20	# Step 1: Add Helm repositories	COMMENT
LOW⚡	deploy/observability/setup-monitoring.sh	27	# Step 2: Update Helm repositories	COMMENT
LOW⚡	deploy/observability/setup-monitoring.sh	32	# Step 3: Install kube-prometheus-stack	COMMENT
LOW⚡	deploy/observability/setup-monitoring.sh	67	# Step 4: Wait for pods to be ready	COMMENT
LOW⚡	deploy/observability/setup-monitoring.sh	73	# Step 5: Verify installation	COMMENT
LOW⚡	deploy/observability/setup-monitoring.sh	78	# Step 6: Update Dynamo operator with Prometheus endpoint	COMMENT
LOW	deploy/observability/setup-monitoring.sh	102	# Step 7: Configure DCGM custom metrics for NVLink profiling	COMMENT
LOW	deploy/observability/setup-monitoring.sh	123	# Step 8: Verify DCGM ServiceMonitor exists (created by GPU operator during cluster setup)	COMMENT
LOW⚡	deploy/observability/setup-monitoring.sh	136	# Step 9: Deploy Grafana dashboard ConfigMap	COMMENT
LOW⚡	deploy/observability/setup-monitoring.sh	142	# Step 10: Get Grafana credentials	COMMENT
LOW	deploy/operator/test/e2e/dgdr/dgdr_test.go	257	// Step 1: Profile without deploying	COMMENT
LOW	deploy/operator/test/e2e/dgdr/dgdr_test.go	276	// Step 2: Verify the output ConfigMap is accessible after profiling	COMMENT
LOW⚡	…loy/operator/test/e2e/dgdr/dgdr_lifecycle_spec_test.go	107	// Step 1: Create DGDR	COMMENT
LOW⚡	…loy/operator/test/e2e/dgdr/dgdr_lifecycle_spec_test.go	115	// Step 2: Wait for profiling to complete (DGDR reaches Ready or later)	COMMENT
LOW⚡	…loy/operator/test/e2e/dgdr/dgdr_lifecycle_spec_test.go	125	// Step 3: Verify ConfigMap output	COMMENT
LOW	…loy/operator/test/e2e/dgdr/dgdr_lifecycle_spec_test.go	147	// Step 4: If autoApply, wait for deployment	COMMENT
LOW	…loy/operator/test/e2e/dgdr/dgdr_lifecycle_spec_test.go	161	// Step 5: Verify DGD reaches successful state	COMMENT
LOW⚡	…loy/operator/test/e2e/dgdr/dgdr_lifecycle_spec_test.go	176	// Step 6: Verify actual pod readiness (independent of DGD status)	COMMENT
LOW⚡	…loy/operator/test/e2e/dgdr/dgdr_lifecycle_spec_test.go	180	// Step 7: Verify individual DGD services	COMMENT
LOW⚡	…loy/operator/test/e2e/dgdr/dgdr_lifecycle_spec_test.go	186	// Step 8: Inference smoke test	COMMENT
LOW	deploy/operator/internal/dynamo/backend_vllm.go	316	// Step 1: Check explicit override	COMMENT
LOW	deploy/operator/internal/dynamo/backend_vllm.go	331	// Step 2: Check operator origin version gate	COMMENT
LOW	…ontroller/dynamocomponentdeployment_controller_test.go	1600	// Step 1: Create the deployment with 1 replica	COMMENT
LOW	…ontroller/dynamocomponentdeployment_controller_test.go	1614	// Step 2: Manually update the deployment to 2 replicas (simulating manual edit)	COMMENT
LOW⚡	…ontroller/dynamocomponentdeployment_controller_test.go	1632	// Step 3: Call createOrUpdateOrDeleteDeployments again - it should reconcile back to 1 replica	COMMENT
LOW⚡	…ontroller/dynamocomponentdeployment_controller_test.go	1638	// Step 4: Verify the deployment was reconciled back to 1 replica	COMMENT
LOW⚡	…ontroller/dynamocomponentdeployment_controller_test.go	1645	// Step 5: Call createOrUpdateOrDeleteDeployments again - it should not be modified	COMMENT
LOW	…/api/v1alpha1/dynamographdeployment_conversion_test.go	1887	// Step 1: first apply. Server ConvertsFrom the user's v1beta1 to	COMMENT
LOW	…/api/v1alpha1/dynamographdeployment_conversion_test.go	1902	// Step 2: second apply. Server reads stored, converts to v1beta1 so	COMMENT
LOW⚡	…/api/v1alpha1/dynamographdeployment_conversion_test.go	1913	// Step 3: apply JSON merge patch. Arrays are replaced wholesale.	COMMENT
LOW⚡	…/api/v1alpha1/dynamographdeployment_conversion_test.go	1923	// Step 4: server converts back to v1alpha1 for storage. This is	COMMENT
LOW⚡	…/api/v1alpha1/dynamographdeployment_conversion_test.go	1930	// Step 5: apiserver generation-bump check.	COMMENT
LOW	…/api/v1alpha1/dynamographdeployment_conversion_test.go	1948	// Step 6: mimic the apiserver's unstructured-level comparison. The	COMMENT
LOW⚡	…alpha1/dynamographdeploymentrequest_conversion_test.go	260	// Step 1: v1alpha1 → v1beta1	COMMENT
LOW⚡	…alpha1/dynamographdeploymentrequest_conversion_test.go	266	// Step 2: v1beta1 → v1alpha1	COMMENT
LOW⚡	…alpha1/dynamographdeploymentrequest_conversion_test.go	319	// Step 1: v1beta1 → v1alpha1	COMMENT
LOW⚡	…alpha1/dynamographdeploymentrequest_conversion_test.go	325	// Step 2: v1alpha1 → v1beta1	COMMENT
LOW⚡	…ploy/operator/api/scripts/generate_pydantic_from_go.py	302	# Step 1: find the index just after the last non-comment line before the block	COMMENT
LOW⚡	…ploy/operator/api/scripts/generate_pydantic_from_go.py	310	# Step 2: forward scan — collect the first paragraph, skip markers	COMMENT
LOW⚡	tests/utils/managed_process.py	571	# Step 3: poll for all process groups to exit.	COMMENT
LOW	tests/utils/managed_process.py	532	# Step 1: Snapshot all process groups before signaling. Only self.proc runs	COMMENT
LOW	tests/utils/managed_process.py	552	# Step 2: SIGTERM all snapshotted process groups (graceful shutdown).	COMMENT
LOW	tests/utils/managed_process.py	594	# Step 4: SIGKILL all snapshotted process groups to ensure nothing	COMMENT
LOW⚡	tests/fault_tolerance/test_vllm_health_check.py	151	# Step 1: Start the frontend	COMMENT
LOW⚡	tests/fault_tolerance/test_vllm_health_check.py	156	# Step 2: Start a worker	COMMENT
LOW⚡	tests/fault_tolerance/test_vllm_health_check.py	163	# Step 3: Send a test request to prove the worker is live.	COMMENT
LOW⚡	tests/fault_tolerance/test_vllm_health_check.py	168	# Step 4: Find and kill vLLM engine processes to force the EngineDeadError condition.	COMMENT
LOW⚡	tests/fault_tolerance/test_vllm_health_check.py	182	# Step 5: Send a request triggering the handler to shutdown everything.	COMMENT
LOW⚡	tests/fault_tolerance/test_vllm_health_check.py	186	# Step 6: Ensure the worker process has been stopped as a result of the EngineDeadError condition.	COMMENT
LOW⚡	tests/fault_tolerance/test_vllm_health_check.py	210	# Step 1: Start the frontend	COMMENT
LOW⚡	tests/fault_tolerance/test_vllm_health_check.py	215	# Step 2: Start a worker	COMMENT
LOW⚡	tests/fault_tolerance/test_vllm_health_check.py	222	# Step 3: Send a test request to prove the worker is live.	COMMENT
LOW⚡	tests/fault_tolerance/test_vllm_health_check.py	227	# Step 4: Find and kill vLLM engine processes to force the EngineDeadError condition.	COMMENT
LOW	tests/fault_tolerance/test_vllm_health_check.py	241	# Step 5: Ensure the worker process has been stopped as a result of the EngineDeadError condition.	COMMENT
LOW⚡	tests/fault_tolerance/cancellation/test_vllm.py	272	# Step 1: Start the frontend (allocates its own frontend_port)	COMMENT
283 more matches not shown…

Modern Structural Boilerplate456 hits · 466 pts

Severity	File	Line	Snippet	Context
LOW	tests/frontend/test_vllm_prepost_integration.py	20	logger = logging.getLogger(__name__)	CODE
LOW	tests/frontend/conftest.py	23	logger = logging.getLogger(__name__)	CODE
LOW	tests/frontend/test_completion_mocker_engine.py	22	logger = logging.getLogger(__name__)	CODE
LOW	tests/frontend/test_vllm.py	27	logger = logging.getLogger(__name__)	CODE
LOW	tests/frontend/test_request_tracing_logs.py	32	logger = logging.getLogger(__name__)	CODE
LOW	tests/frontend/test_realtime_python_bridge.py	32	logger = logging.getLogger(__name__)	CODE
LOW	tests/frontend/test_realtime_omni_bridge.py	34	logger = logging.getLogger(__name__)	CODE
LOW	tests/frontend/test_self_host_metadata.py	28	logger = logging.getLogger(__name__)	CODE
LOW	tests/frontend/test_prompt_embeds.py	39	logger = logging.getLogger(__name__)	CODE
LOW	tests/frontend/test_frontend_api_surface_compliance.py	58	logger = logging.getLogger(__name__)	CODE
LOW	tests/frontend/test_tool_calling_sglang.py	41	logger = logging.getLogger(__name__)	CODE
LOW	tests/frontend/grpc/test_tensor_parameters.py	27	logger = logging.getLogger(__name__)	CODE
LOW	tests/frontend/grpc/test_tensor_mocker_engine.py	37	logger = logging.getLogger(__name__)	CODE
LOW	tests/gpu_memory_service/conftest.py	18	logger = logging.getLogger(__name__)	CODE
LOW	tests/gpu_memory_service/test_shadow_failover.py	43	logger = logging.getLogger(__name__)	CODE
LOW	tests/gpu_memory_service/flow_assertions.py	16	logger = logging.getLogger(__name__)	CODE
LOW	tests/gpu_memory_service/common/runtime.py	24	logger = logging.getLogger(__name__)	CODE
LOW	tests/deploy/test_deploy.py	30	logger = logging.getLogger(__name__)	CODE
LOW	tests/deploy/test_dynamocheckpoint.py	26	logger = logging.getLogger(__name__)	CODE
LOW	tests/basic/test_autodeploy_backend.py	18	logger = logging.getLogger(__name__)	CODE
LOW	tests/utils/device.py	13	logger = logging.getLogger(__name__)	CODE
LOW	tests/utils/client.py	13	logger = logging.getLogger(__name__)	CODE
LOW	tests/utils/profile_pytest.py	68	logger = logging.getLogger(__name__)	CODE
LOW	tests/utils/engine_process.py	17	logger = logging.getLogger(__name__)	CODE
LOW	tests/utils/test_mock_gpu_alloc.py	26	logger = logging.getLogger(__name__)	CODE
LOW	tests/utils/managed_deployment.py	275	def _set_args(self, tokens: list[str]) -> None:	CODE
LOW	tests/utils/payloads.py	35	logger = logging.getLogger(__name__)	CODE
LOW	tests/fault_tolerance/test_vllm_health_check.py	18	logger = logging.getLogger(__name__)	CODE
LOW	tests/fault_tolerance/test_canary_rank_pause.py	30	logger = logging.getLogger(__name__)	CODE
LOW	tests/fault_tolerance/cancellation/test_vllm.py	37	logger = logging.getLogger(__name__)	CODE
LOW	tests/fault_tolerance/cancellation/test_trtllm.py	33	logger = logging.getLogger(__name__)	CODE
LOW	tests/fault_tolerance/cancellation/test_sglang.py	31	logger = logging.getLogger(__name__)	CODE
LOW	tests/fault_tolerance/cancellation/utils.py	20	logger = logging.getLogger(__name__)	CODE
LOW	…_service/agents/gpu_fault_injector/gpu_xid_injector.py	76	logger = logging.getLogger(__name__)	CODE
LOW	…t_injection_service/agents/gpu_fault_injector/agent.py	41	logger = logging.getLogger(__name__)	CODE
LOW⚡	…e/hardware/fault_injection_service/api_service/main.py	32	logger = logging.getLogger(__name__)	CODE
LOW	…e/hardware/fault_injection_service/helpers/__init__.py	11	__all__ = [	CODE
LOW	…ware/fault_injection_service/helpers/k8s_operations.py	19	logger = logging.getLogger(__name__)	CODE
LOW	…dware/fault_injection_service/helpers/gpu_discovery.py	17	logger = logging.getLogger(__name__)	CODE
LOW	tests/fault_tolerance/deploy/scenarios.py	32	logger = logging.getLogger(__name__)	CODE
LOW	tests/fault_tolerance/deploy/k8s_utils.py	28	logger = logging.getLogger(__name__)	CODE
LOW⚡	tests/fault_tolerance/deploy/checkers.py	38	logger = logging.getLogger(__name__)	CODE
LOW	tests/fault_tolerance/deploy/base_checker.py	29	logger = logging.getLogger(__name__)	CODE
LOW	tests/fault_tolerance/deploy/checker_factory.py	39	logger = logging.getLogger(__name__)	CODE
LOW	tests/fault_tolerance/deploy/validation_checks.py	28	logger = logging.getLogger(__name__)	CODE
LOW	tests/fault_tolerance/etcd_ha/test_vllm.py	32	logger = logging.getLogger(__name__)	CODE
LOW	tests/fault_tolerance/etcd_ha/test_trtllm.py	23	logger = logging.getLogger(__name__)	CODE
LOW	tests/fault_tolerance/etcd_ha/test_sglang.py	27	logger = logging.getLogger(__name__)	CODE
LOW	tests/fault_tolerance/etcd_ha/utils.py	25	logger = logging.getLogger(__name__)	CODE
LOW	tests/fault_tolerance/migration/test_vllm.py	27	logger = logging.getLogger(__name__)	CODE
LOW	tests/fault_tolerance/migration/test_trtllm.py	26	logger = logging.getLogger(__name__)	CODE
LOW	tests/fault_tolerance/migration/test_sglang.py	26	logger = logging.getLogger(__name__)	CODE
LOW	tests/fault_tolerance/migration/utils.py	19	logger = logging.getLogger(__name__)	CODE
LOW	tests/kvbm_integration/test_consolidator_router_e2e.py	55	logger = logging.getLogger(__name__)	CODE
LOW	tests/kvbm_integration/test_cuda_graph.py	31	logger = logging.getLogger(__name__)	CODE
LOW	tests/vllm_self_benchmark/test_self_benchmark_gpu.py	73	logger = logging.getLogger(__name__)	CODE
LOW	tests/serve/test_vllm.py	52	logger = logging.getLogger(__name__)	CODE
LOW	tests/serve/test_sample.py	23	logger = logging.getLogger(__name__)	CODE
LOW	tests/serve/test_trtllm.py	34	logger = logging.getLogger(__name__)	CODE
LOW	tests/serve/test_sglang.py	48	logger = logging.getLogger(__name__)	CODE
396 more matches not shown…

Deep Nesting428 hits · 356 pts

Severity	File	Line	Context
LOW	deploy/power-agent/power_agent.py	874	CODE
LOW	deploy/power-agent/tests/e2e_actuator_parity.py	146	CODE
LOW	deploy/utils/dynamo_deployment.py	310	CODE
LOW	deploy/utils/dynamo_deployment.py	369	CODE
LOW	…ploy/operator/api/scripts/generate_pydantic_from_go.py	203	CODE
LOW	…ploy/operator/api/scripts/generate_pydantic_from_go.py	265	CODE
LOW	…ploy/operator/api/scripts/generate_pydantic_from_go.py	334	CODE
LOW	…ploy/operator/api/scripts/generate_pydantic_from_go.py	460	CODE
LOW	tests/conftest.py	651	CODE
LOW	tests/conftest.py	1147	CODE
LOW	tests/conftest.py	1174	CODE
LOW	tests/frontend/test_vllm_prepost_integration.py	143	CODE
LOW	tests/frontend/conftest.py	79	CODE
LOW	tests/frontend/test_request_tracing_logs.py	242	CODE
LOW	tests/frontend/test_request_tracing_logs.py	594	CODE
LOW	tests/frontend/test_realtime_python_bridge.py	164	CODE
LOW	tests/frontend/test_unified_worker_otlp_export.py	156	CODE
LOW	tests/frontend/test_unified_worker_otlp_export.py	234	CODE
LOW	tests/frontend/test_unified_worker_otlp_export.py	375	CODE
LOW	tests/frontend/test_prepost_mistral.py	528	CODE
LOW	tests/frontend/test_realtime_omni_bridge.py	144	CODE
LOW	tests/frontend/test_prepost.py	1444	CODE
LOW	tests/frontend/test_frontend_api_surface_compliance.py	171	CODE
LOW	tests/frontend/test_unified_worker_tracing_smoke.py	72	CODE
LOW	tests/frontend/test_tool_calling_sglang.py	512	CODE
LOW	tests/frontend/grpc/test_tensor_mocker_engine.py	174	CODE
LOW	tests/frontend/grpc/echo_tensor_worker.py	57	CODE
LOW	tests/deploy/test_dynamocheckpoint.py	230	CODE
LOW	tests/basic/test_cuda_version_consistency.py	84	CODE
LOW	tests/utils/client.py	26	CODE
LOW	tests/utils/client.py	49	CODE
LOW	tests/utils/client.py	155	CODE
LOW	tests/utils/profile_pytest.py	118	CODE
LOW	tests/utils/profile_pytest.py	832	CODE
LOW	tests/utils/test_managed_process_teardown.py	164	CODE
LOW	tests/utils/pytest_parallel_gpu.py	448	CODE
LOW	tests/utils/pytest_parallel_gpu.py	677	CODE
LOW	tests/utils/managed_deployment.py	888	CODE
LOW	tests/utils/managed_deployment.py	1007	CODE
LOW	tests/utils/managed_deployment.py	1077	CODE
LOW	tests/utils/managed_deployment.py	1508	CODE
LOW	tests/utils/managed_deployment.py	1603	CODE
LOW	tests/utils/managed_process.py	247	CODE
LOW	tests/utils/managed_process.py	325	CODE
LOW	tests/utils/managed_process.py	498	CODE
LOW	tests/utils/managed_process.py	620	CODE
LOW	tests/utils/managed_process.py	683	CODE
LOW	tests/utils/managed_process.py	830	CODE
LOW	tests/utils/payloads.py	225	CODE
LOW	tests/utils/payloads.py	735	CODE
LOW	tests/utils/payloads.py	907	CODE
LOW	tests/utils/payloads.py	1031	CODE
LOW	tests/utils/payloads.py	1645	CODE
LOW	tests/utils/port_utils.py	93	CODE
LOW	tests/utils/port_utils.py	205	CODE
LOW	tests/utils/payload_builder.py	303	CODE
LOW	tests/wheels/smoke_install.py	219	CODE
LOW	tests/wheels/smoke_install.py	235	CODE
LOW	tests/fault_tolerance/test_vllm_health_check.py	142	CODE
LOW	tests/fault_tolerance/test_vllm_health_check.py	199	CODE
368 more matches not shown…

Cross-File Repetition52 hits · 260 pts

Severity	File	Snippet	Context
HIGH	…onents/operator/crds/nvidia.com_dynamocheckpoints.yaml	type: string resourcefieldref: description: \|- selects a resource of the container: only resources limits and requests (	STRING
HIGH	…tor/crds/nvidia.com_dynamographdeploymentrequests.yaml	type: string resourcefieldref: description: \|- selects a resource of the container: only resources limits and requests (	STRING
HIGH	…tor/config/crd/bases/nvidia.com_dynamocheckpoints.yaml	type: string resourcefieldref: description: \|- selects a resource of the container: only resources limits and requests (	STRING
HIGH	…rd/bases/nvidia.com_dynamographdeploymentrequests.yaml	type: string resourcefieldref: description: \|- selects a resource of the container: only resources limits and requests (	STRING
HIGH	tests/fault_tolerance/etcd_ha/test_vllm.py	test etcd high availability with repeated node failures and recoveries for trt-llm in disaggregated mode. this test: 1.	STRING
HIGH	tests/fault_tolerance/etcd_ha/test_trtllm.py	test etcd high availability with repeated node failures and recoveries for trt-llm in disaggregated mode. this test: 1.	STRING
HIGH	tests/fault_tolerance/etcd_ha/test_sglang.py	test etcd high availability with repeated node failures and recoveries for trt-llm in disaggregated mode. this test: 1.	STRING
HIGH	tests/fault_tolerance/migration/test_vllm.py	end-to-end test for aggregated worker request migration. parameters: immediate_kill: true for abrupt kill (sigkill), fal	STRING
HIGH	tests/fault_tolerance/migration/test_trtllm.py	end-to-end test for aggregated worker request migration. parameters: immediate_kill: true for abrupt kill (sigkill), fal	STRING
HIGH	tests/fault_tolerance/migration/test_sglang.py	end-to-end test for aggregated worker request migration. parameters: immediate_kill: true for abrupt kill (sigkill), fal	STRING
HIGH	tests/fault_tolerance/migration/test_vllm.py	end-to-end test for decode worker request migration in disaggregated mode. setup: 1 prefill worker + 2 decode workers pa	STRING
HIGH	tests/fault_tolerance/migration/test_trtllm.py	end-to-end test for decode worker request migration in disaggregated mode. setup: 1 prefill worker + 2 decode workers pa	STRING
HIGH	tests/fault_tolerance/migration/test_sglang.py	end-to-end test for decode worker request migration in disaggregated mode. setup: 1 prefill worker + 2 decode workers pa	STRING
HIGH	tests/kvbm_integration/test_determinism_agg.py	create determinism tester bound to the running server's base url.	STRING
HIGH	tests/kvbm_integration/common.py	create determinism tester bound to the running server's base url.	STRING
HIGH	tests/kvbm_integration/test_determinism_disagg.py	create determinism tester bound to the running server's base url.	STRING
HIGH	tests/kvbm_integration/test_determinism_agg.py	test determinism across cache reset: run test with warmup, reset cache, run again without warmup.	STRING
HIGH	tests/kvbm_integration/common.py	test determinism across cache reset: run test with warmup, reset cache, run again without warmup.	STRING
HIGH	tests/kvbm_integration/test_determinism_disagg.py	test determinism across cache reset: run test with warmup, reset cache, run again without warmup.	STRING
HIGH	tests/serve/test_vllm.py	test lora inference with aggregated sglang deployment. this test: 1. uses minio fixture to provide s3-compatible storage	STRING
HIGH	tests/serve/test_vllm_xpu.py	test lora inference with aggregated sglang deployment. this test: 1. uses minio fixture to provide s3-compatible storage	STRING
HIGH	tests/serve/test_sglang.py	test lora inference with aggregated sglang deployment. this test: 1. uses minio fixture to provide s3-compatible storage	STRING
HIGH	tests/router/test_router_e2e_with_sglang.py	validate kv cache prefix reuse with vllm by sending progressive requests with overlapping prefixes. same flow as test_ro	STRING
HIGH	tests/router/test_router_e2e_with_trtllm.py	validate kv cache prefix reuse with vllm by sending progressive requests with overlapping prefixes. same flow as test_ro	STRING
HIGH	tests/router/test_router_e2e_with_vllm.py	validate kv cache prefix reuse with vllm by sending progressive requests with overlapping prefixes. same flow as test_ro	STRING
HIGH	components/src/dynamo/vllm/llm_engine.py	tensorrt-llm llmengine implementation for the unified backend. see dynamo/common/backend/readme.md for architecture, res	STRING
HIGH	components/src/dynamo/sglang/llm_engine.py	tensorrt-llm llmengine implementation for the unified backend. see dynamo/common/backend/readme.md for architecture, res	STRING
HIGH	components/src/dynamo/trtllm/llm_engine.py	tensorrt-llm llmengine implementation for the unified backend. see dynamo/common/backend/readme.md for architecture, res	STRING
HIGH	components/src/dynamo/vllm/tests/test_vllm_unit.py	test that invalid file path raises filenotfounderror.	STRING
HIGH	components/src/dynamo/sglang/tests/test_sglang_unit.py	test that invalid file path raises filenotfounderror.	STRING
HIGH	components/src/dynamo/trtllm/tests/test_trtllm_unit.py	test that invalid file path raises filenotfounderror.	STRING
HIGH	components/src/dynamo/vllm/tests/test_vllm_unit.py	test that environment variables in paths are expanded by python code.	STRING
HIGH	components/src/dynamo/sglang/tests/test_sglang_unit.py	test that environment variables in paths are expanded by python code.	STRING
HIGH	components/src/dynamo/trtllm/tests/test_trtllm_unit.py	test that environment variables in paths are expanded by python code.	STRING
HIGH	…s/src/dynamo/vllm/tests/test_vllm_trace_propagation.py	async iterator that yields nothing. the unreachable ``yield`` is what makes this function an async generator rather than	STRING
HIGH	…c/dynamo/sglang/tests/test_sglang_trace_propagation.py	async iterator that yields nothing. the unreachable ``yield`` is what makes this function an async generator rather than	STRING
HIGH	…c/dynamo/trtllm/tests/test_trtllm_trace_propagation.py	async iterator that yields nothing. the unreachable ``yield`` is what makes this function an async generator rather than	STRING
HIGH	…rc/dynamo/common/tests/multimodal/test_audio_loader.py	wiring smoke: videoloader plumbs ``url_policy`` to the validator. validator behavior is covered in ``test_url_validator.	STRING
HIGH	…rc/dynamo/common/tests/multimodal/test_image_loader.py	wiring smoke: videoloader plumbs ``url_policy`` to the validator. validator behavior is covered in ``test_url_validator.	STRING
HIGH	…rc/dynamo/common/tests/multimodal/test_video_loader.py	wiring smoke: videoloader plumbs ``url_policy`` to the validator. validator behavior is covered in ``test_url_validator.	STRING
HIGH	…c/dynamo/profiler/tests/unit/test_profiler_protocol.py	set dgdr_name so _validate_dgd_service_name_lengths runs in tests.	STRING
HIGH	…dynamo/profiler/tests/unit/test_helpers_profile_sla.py	set dgdr_name so _validate_dgd_service_name_lengths runs in tests.	STRING
HIGH	…mo/profiler/tests/integration/test_profile_sla_dgdr.py	set dgdr_name so _validate_dgd_service_name_lengths runs in tests.	STRING
HIGH	…nts/src/dynamo/profiler/utils/config_modifiers/vllm.py	update container image for all dgd services (frontend, planner, workers).	STRING
HIGH	…s/src/dynamo/profiler/utils/config_modifiers/trtllm.py	update container image for all dgd services (frontend, planner, workers).	STRING
HIGH	…s/src/dynamo/profiler/utils/config_modifiers/sglang.py	update container image for all dgd services (frontend, planner, workers).	STRING
HIGH	lib/bindings/python/examples/hello_world/server.py	create and serve the `generate` endpoint using the distributed runtime. multiple endpoints can be served from a single w	STRING
HIGH	lib/bindings/python/examples/error_handling/server.py	create and serve the `generate` endpoint using the distributed runtime. multiple endpoints can be served from a single w	STRING
HIGH	lib/bindings/python/examples/typed/server.py	create and serve the `generate` endpoint using the distributed runtime. multiple endpoints can be served from a single w	STRING
HIGH	lib/bindings/python/examples/hello_world/client.py	instantiate a `backend` client and call the `generate` endpoint	STRING
HIGH	lib/bindings/python/examples/error_handling/client.py	instantiate a `backend` client and call the `generate` endpoint	STRING
HIGH	lib/bindings/python/examples/typed/client.py	instantiate a `backend` client and call the `generate` endpoint	STRING

AI Slop Vocabulary97 hits · 256 pts

Severity	File	Line	Snippet	Context
MEDIUM	deploy/helm/charts/power-agent/templates/dev-pod.yaml	7	# In-cluster dev-iteration harness. Renders a single Pod that mounts	COMMENT
MEDIUM	…perator/internal/namespace_scope/lease_manager_test.go	263	// Poll for renewal with timeout (more robust than fixed sleep)	COMMENT
MEDIUM	tests/utils/collection_env_guard.py	46	# Nothing should mutate a watched var during collection -- legitimate harness	COMMENT
MEDIUM	tests/utils/managed_process.py	457	# Capture the child's process group id for robust cleanup even if parent shell exits	COMMENT
MEDIUM	tests/utils/managed_process.py	484	# Capture the child's process group id for robust cleanup even if parent shell exits	COMMENT
MEDIUM	tests/utils/payloads.py	1319	# by the harness). Each indexed worker must have observed >0 delta	COMMENT
MEDIUM	tests/utils/payloads.py	1323	# config-construction time because the harness assigns dynamic ports.	COMMENT
MEDIUM	tests/fault_tolerance/deploy/scenarios.py	489	"""Log comprehensive pod diagnostics including process list, GPU info, and nvidia-smi."""	STRING
MEDIUM	tests/kvbm_integration/common.py	359	"""Perform comprehensive server warmup with all test prompts."""	STRING
MEDIUM	tests/kvbm_integration/common.py	413	"""Run the test iterations with comprehensive warmup."""	STRING
MEDIUM	tests/serve/test_vllm.py	738	# Pooling models do not use a KV cache, but the test harness still	COMMENT
MEDIUM	tests/serve/test_vllm.py	866	# ready-on-liveness so the harness's worker health check passes; the	COMMENT
MEDIUM	tests/serve/test_sglang.py	923	# the harness's worker health check passes (canary-having workers still	COMMENT
MEDIUM	tests/serve/common.py	50	# subclasses (e.g. ChatPayloadWithLogprobs) and is robust to ad-hoc `endpoint`	COMMENT
MEDIUM	tests/router/test_router_e2e_with_vllm.py	160	# harness launches internal-LB DP (only --data-parallel-size, no	COMMENT
MEDIUM	tests/router/test_router_e2e_with_vllm_xpu.py	195	# health_check_ports=[kv_event_port]: the harness will poll	COMMENT
MEDIUM	.agents/skills/dynamo-frontend-benchmark/scripts/env.sh	5	# Shared environment for the Dynamo frontend benchmark harness.	COMMENT
MEDIUM	container/compliance/generators/dpkg.py	214	# Detecting these references is the highest-leverage lever for cutting UNKNOWNs.	COMMENT
MEDIUM	components/src/dynamo/frontend/sglang_prepost.py	977	# Full text accumulator for robust finish-time re-parse.	COMMENT
LOW	components/src/dynamo/common/utils/output_modalities.py	60	# For now, we ignore model repo and just use cli input to determine output modalities.	COMMENT
LOW	components/src/dynamo/trtllm/workers/llm_worker.py	534	# So for now, we just set the parsers from the config	COMMENT
MEDIUM	examples/custom_encoder/launch/agg_custom.sh	45	# Precedence: explicit DYN_WORKER_GPU/--gpu > harness-set CUDA_VISIBLE_DEVICES	COMMENT
MEDIUM	examples/custom_encoder/launch/agg_custom.sh	91	# VRAM sizing: honors the profiler/test-harness override	COMMENT
MEDIUM⚡	…ds/vllm/launch/agg_multimodal_router_chat_processor.sh	37	# Honor DYN_HTTP_PORT (set by the tests/serve harness for dynamic	COMMENT
MEDIUM	…ds/vllm/launch/agg_multimodal_router_chat_processor.sh	80	# The serve-test harness invokes scripts as `bash <script> --model NAME ...`,	COMMENT
MEDIUM	examples/backends/vllm/launch/disagg_same_gpu.sh	27	# Inherit GPU from parent (profiler/test harness sets CUDA_VISIBLE_DEVICES);	COMMENT
MEDIUM⚡	…launch/xpu/agg_multimodal_router_chat_processor_xpu.sh	37	# Honor DYN_HTTP_PORT (set by the tests/serve harness for dynamic	COMMENT
MEDIUM	…launch/xpu/agg_multimodal_router_chat_processor_xpu.sh	80	# The serve-test harness invokes scripts as `bash <script> --model NAME ...`,	COMMENT
MEDIUM	…ples/backends/sample/launch/multimodal_smoke_client.py	175	# a fresh timeout that can exceed the outer test harness deadline.	COMMENT
MEDIUM	examples/backends/sglang/launch/diffusion_llada.sh	25	# Manual HTTP_PORT override, else the harness-assigned DYN_HTTP_PORT, else the default.	COMMENT
MEDIUM	examples/backends/sglang/launch/agg_vision.sh	85	# Profiler/test-harness override: when _PROFILE_OVERRIDE_SGLANG_MAX_TOTAL_TOKENS is	COMMENT
MEDIUM	…amples/backends/sglang/launch/agg_multimodal_router.sh	23	# Honor DYN_HTTP_PORT (set by tests/serve harness for dynamic port allocation)	COMMENT
MEDIUM	…amples/backends/sglang/launch/agg_multimodal_router.sh	176	# the harness may have applied (instead of the default formula).	COMMENT
MEDIUM	examples/backends/trtllm/launch/disagg_e_pd.sh	34	# Profiler/test-harness override applied to the KV-cache-bearing PD worker.	COMMENT
MEDIUM	…s/trtllm/launch/epd_multimodal_image_and_embeddings.sh	30	# Profiler/test-harness override applied to KV-cache-bearing workers (prefill, decode).	COMMENT
MEDIUM	examples/backends/trtllm/launch/disagg_multimodal.sh	21	# Profiler/test-harness override applied to KV-cache-bearing workers (prefill, decode).	COMMENT
MEDIUM	examples/backends/trtllm/launch/agg_multimodal.sh	18	# Profiler/test-harness override: when _PROFILE_OVERRIDE_TRTLLM_MAX_TOTAL_TOKENS or	COMMENT
MEDIUM	examples/backends/trtllm/launch/agg_router_approx.sh	18	# Profiler/test-harness override applied to the KV-cache-bearing worker.	COMMENT
MEDIUM	…amples/backends/trtllm/launch/agg_multimodal_router.sh	28	# Profiler/test-harness override: KvCacheConfig JSON when env var is set, empty otherwise.	COMMENT
MEDIUM⚡	examples/common/launch_utils.sh	59	# harness signalling the process group) otherwise interrupts wait -n and	COMMENT
MEDIUM⚡	examples/common/launch_utils.sh	70	# subreaper (pytest in CI), and the harness's pgid liveness check	COMMENT
MEDIUM	.github/filters.yaml	167	# Standalone CRD apiVersion converter: not used by the deploy-test harness and	COMMENT
MEDIUM	.github/actions/docker-remote-build/action.yml	286	# Output single comprehensive JSON with all build stages	COMMENT
MEDIUM	.github/actions/docker-remote-build/action.yml	334	# Upload comprehensive build metrics as artifact	COMMENT
LOW	.github/actions/bootstrap-buildkit/action.yml	90	# If the builder already exists, these commands will just set the right configurations to use it	COMMENT
MEDIUM	.github/actions/docker-build/action.yml	273	# Output single comprehensive JSON with all build stages	COMMENT
MEDIUM	.github/actions/docker-build/action.yml	321	# Upload comprehensive build metrics as artifact	COMMENT
MEDIUM	lib/llm/tests/kv_manager.rs	203	// /// and hold them until a RDMA GET COMPLETION notification is received. Essentially, the task which	COMMENT
MEDIUM	…ests/data/deepseek-v3.2/test_input_search_wo_date.json	243	"content": "[4] www.annualreports.com (https://www.annualreports.com/HostedData/AnnualReportArchive/d/defend	CODE
MEDIUM	…ests/data/deepseek-v3.2/test_input_search_wo_date.json	283	"content": "[6] www.annualreports.com (https://www.annualreports.com/HostedData/AnnualReportArchive/d/defend	CODE
MEDIUM	…ests/data/deepseek-v3.2/test_input_search_wo_date.json	323	"content": "[8] www.annualreports.com (https://www.annualreports.com/HostedData/AnnualReportArchive/d/defend	CODE
MEDIUM	…ests/data/deepseek-v3.2/test_input_search_wo_date.json	423	"content": "[13] As of December 5, 2021, Kenya had received a total of 16,201,670 vaccines (Search_Results/A	CODE
MEDIUM	…ests/data/deepseek-v3.2/test_input_search_wo_date.json	443	"content": "[14] www.worldbank.org (https://www.worldbank.org/en/news/press-release/2021/12/14/kenya-s-econo	CODE
MEDIUM⚡	…tests/data/deepseek-v3.2/test_input_search_w_date.json	199	"content": "[2] (https://ar5iv.org/html/2510.19631v1)\nviewing lines [0 - 119] of 983\n\nL0: \nL1: URL:	CODE
MEDIUM⚡	…tests/data/deepseek-v3.2/test_input_search_w_date.json	199	"content": "[2] (https://ar5iv.org/html/2510.19631v1)\nviewing lines [0 - 119] of 983\n\nL0: \nL1: URL:	CODE
MEDIUM⚡	…tests/data/deepseek-v3.2/test_input_search_w_date.json	199	"content": "[2] (https://ar5iv.org/html/2510.19631v1)\nviewing lines [0 - 119] of 983\n\nL0: \nL1: URL:	CODE
MEDIUM⚡	…tests/data/deepseek-v3.2/test_input_search_w_date.json	431	"content": "[13] BEIR benchmark (Search_Results/BEIR benchmark)\nviewing lines [0 - 37] of 37\n\nL0: \nL	CODE
MEDIUM⚡	…tests/data/deepseek-v3.2/test_input_search_w_date.json	436	"content": "[14] MIRACL benchmark (Search_Results/MIRACL benchmark)\nviewing lines [0 - 41] of 41\n\nL0:	CODE
MEDIUM⚡	…tests/data/deepseek-v3.2/test_input_search_w_date.json	441	"content": "[15] Zero-shot IR benchmark (Search_Results/Zero-shot IR benchmark)\n**viewing lines [0 - 40] of	CODE
MEDIUM⚡	…tests/data/deepseek-v3.2/test_input_search_w_date.json	461	"content": "[19] Mind2Web benchmark (Search_Results/Mind2Web benchmark)\nviewing lines [0 - 40] of 40\n\	CODE
37 more matches not shown…

Docstring Block Structure49 hits · 245 pts

Severity	File	Line	Snippet	Context
HIGH	tests/frontend/conftest.py	82	Check if the gRPC server process is up and accepting connections. Uses KServe `server_live` rather than `server_rea	STRING
HIGH	tests/deploy/test_deploy.py	58	Validate the structure and content of a chat completion response. Args: response: HTTP response from the ch	STRING
HIGH	tests/utils/client.py	57	Send an HTTP request to the engine with detailed logging. Args: url: The endpoint URL payload:	STRING
HIGH	tests/utils/test_output.py	18	Resolve a test output path to an absolute path. This function ensures test output is written to a dedicated locatio	STRING
HIGH	…_service/agents/gpu_fault_injector/gpu_xid_injector.py	157	Get PCI address for GPU by reading /host/proc/driver/nvidia/gpus/. This method works without nvidia-sm	STRING
HIGH	tests/fault_tolerance/deploy/parse_factory.py	110	Auto-detect and parse test results using the appropriate parser. This function automatically detects whether result	STRING
HIGH	tests/fault_tolerance/deploy/client_factory.py	22	Get the appropriate client function based on client type. This factory function returns the correct client implemen	STRING
HIGH	tests/fault_tolerance/deploy/client_factory.py	95	Get a human-readable description of a client type. Args: client_type: Client type to describe Returns:	STRING
HIGH	tests/kvbm_integration/common.py	1331	Fetch and parse KVBM metrics from the metrics endpoint. Args: port: Metrics server port (required for dynam	STRING
HIGH	tests/router/helper.py	332	Poll the endpoint's discovery client until the expected number of worker instances appear. Args: endpoint:	STRING
HIGH	tests/router/helper.py	373	Wait for workers to be ready and return their instance IDs. Supports mocker and vLLM workers. This function pol	STRING
HIGH	components/src/dynamo/planner/connectors/kubernetes.py	325	Get the GPU counts for prefill and decode components from the deployment. Args: deployment: Optiona	STRING
HIGH	components/src/dynamo/planner/plugins/transport/base.py	60	Invoke a plugin RPC by method name. Args: method: RPC method name. One of ``"Predict"`` / ``"Propos	STRING
HIGH	components/src/dynamo/vllm/envs.py	27	Resolve port from environment variable with validation. Args: env_var: Environment variable name	STRING
HIGH	components/src/dynamo/vllm/handlers.py	2379	Decode base64-encoded prompt embeddings in PyTorch format. Use vllm's safe loader to prevent out-of-bo	STRING
HIGH	components/src/dynamo/vllm/handlers.py	2414	Decode prompt embeddings and create EmbedsPrompt for vLLM. Args: prompt_embeds_base64: Bas	STRING
HIGH	…ts/src/dynamo/vllm/multimodal_utils/embed_assembler.py	46	Build the mixed token-ids/embeds inputs for an aggregated request. Args: token_ids: The full prompt token I	STRING
HIGH	components/src/dynamo/vllm/multimodal_utils/model.py	106	Return the multimodal model family for the given identifier, or `None` if no registered family matches. Resolut	STRING
HIGH	…nents/src/dynamo/vllm/multimodal_utils/encode_utils.py	47	Extract image features using Qwen-style vision encoder. Args: vision_encoder: The vision encoder model	STRING
HIGH	…nents/src/dynamo/vllm/multimodal_utils/encode_utils.py	92	Encode image embeddings using the appropriate model-specific encoder. Args: model_name: The model iden	STRING
HIGH	…nents/src/dynamo/vllm/multimodal_utils/encode_utils.py	136	Get the appropriate vision encoder and projector components for a given model. Args: model_name: The m	STRING
HIGH	…mo/common/memory/multimodal_embedding_cache_manager.py	79	Calculate tensor size in bytes. Args: tensor: Must be a contiguous tensor. Returns:	STRING
HIGH	components/src/dynamo/common/utils/endpoint_types.py	10	Parse comma-separated endpoint types into ModelType flags. Args: endpoint_types_str: Comma-separated list o	STRING
HIGH	components/src/dynamo/common/utils/image_utils.py	20	Encode numpy image to PNG bytes (in-memory). Args: image: Numpy image array of shape (H, W, 3) with uint8 v	STRING
HIGH	components/src/dynamo/common/utils/video_utils.py	86	Convert a list of PIL Images to a numpy array suitable for video encoding. Args: images: List of PIL Image	STRING
HIGH	components/src/dynamo/common/utils/video_utils.py	123	Encode numpy frames to MP4 file. Args: frames: Video frames as numpy array of shape (num_frames, height, wi	STRING
HIGH	components/src/dynamo/common/utils/video_utils.py	185	Encode numpy frames to video bytes (in-memory). Args: frames: Video frames as numpy array of shape (num_fra	STRING
HIGH	components/src/dynamo/common/utils/runtime.py	22	Parse a Dynamo endpoint string into its components. Args: endpoint: Endpoint string in format 'namespace.co	STRING
HIGH	components/src/dynamo/common/utils/topology.py	118	Read topology config from a file-backed source and env vars. The deployment environment injects env vars for topolo	STRING
HIGH	components/src/dynamo/common/config_dump/environment.py	56	Get relevant environment variables based on prefixes. Args: prefixes: List of environment variable pre	STRING
HIGH	…ts/src/dynamo/common/multimodal/async_encoder_cache.py	84	Get from cache or compute with request coalescing. If the key is in cache, returns immediately.	STRING
HIGH	components/src/dynamo/common/multimodal/image_loader.py	248	Load a batch of images from multimodal data items. Supports two paths: 1. Url variant: Downloa	STRING
HIGH	components/src/dynamo/mocker/args.py	72	Resolve --planner-profile-data to an NPZ file path. Handles backward compatibility by accepting either: 1.	STRING
HIGH	components/src/dynamo/sglang/args.py	316	Parse CLI arguments and return combined configuration. Download the model if necessary. Args: args: Com	STRING
HIGH	…c/dynamo/sglang/request_handlers/llm/decode_handler.py	331	Generate response in aggregated or disaggregated mode. Args: request: Request dict with input and s	STRING
HIGH	components/src/dynamo/profiler/utils/profile_common.py	74	Derive the backend worker image from the profiler image. Replaces the image name (the last ``/``-delimited componen	STRING
HIGH	…src/dynamo/profiler/utils/config_modifiers/protocol.py	173	Extract model name and path from worker args. Checks --served-model-name first (API name), then falls	STRING
HIGH	…src/dynamo/profiler/utils/config_modifiers/protocol.py	496	Build a complete DynamoGraphDeployment config by loading a base YAML and injecting pre-computed CLI arg	STRING
HIGH	components/src/dynamo/global_router/pool_selection.py	623	Load Global Router configuration from JSON file. Args: config_path: Path to the JSON configuration fil	STRING
HIGH	components/src/dynamo/trtllm/encode_helper.py	147	Read embeddings from encode worker response using NIXL and reconstruct original format. Args:	STRING
HIGH	…amo/trtllm/request_handlers/base_generative_handler.py	28	Generate response from request. This is the main entry point called by Dynamo's endpoint.serve_endpoint().	STRING
HIGH	…amo/trtllm/request_handlers/diffusion/video_handler.py	76	Parse 'WxH' string to (width, height) tuple. The API accepts size as a string (e.g., "832x480") to match the fo	STRING
HIGH	…amo/trtllm/request_handlers/diffusion/image_handler.py	76	Parse 'WxH' string to (width, height) tuple. The API accepts size as a string (e.g., "832x480") to match the fo	STRING
HIGH	components/src/dynamo/trtllm/multimodal/cuda_ipc.py	17	Extract all embedding tensors from CUDA IPC handles and move to CPU. Runs extraction in a worker thread to avo	STRING
HIGH	…ents/src/dynamo/trtllm/multimodal/embedding_fetcher.py	34	Fetch embeddings from remote encode worker. Args: image_urls: List of image URLs to encode (must not b	STRING
HIGH	…ents/src/dynamo/trtllm/multimodal/embedding_fetcher.py	83	Call encode worker for full EPD flow. Args: request: Request dict encode_client: Client to cal	STRING
HIGH	…mponents/src/dynamo/trtllm/engines/diffusion_engine.py	170	Generate video/image frames from text prompt. This is a synchronous method that should be called from a thread	STRING
HIGH	lib/gpu_memory_service/integrations/sglang/__init__.py	36	Setup GPU Memory Service for SGLang. Validates config and returns the GMSModelLoader class. Patches are applied	STRING
HIGH	…/trtllm_integration/connector/kvbm_connector_worker.py	41	Create a dedicated NCCL communicator for KVBM using MPI for bootstrap. This function creates an NCCL communicator t	STRING

Modern AI Meta-Vocabulary67 hits · 192 pts

Severity	File	Line	Snippet	Context
MEDIUM	deploy/helm/charts/platform/README.md	222	enabled: true # Enables multinode orchestration via PodCliqueSets	CODE
MEDIUM	deploy/helm/charts/platform/README.md	232	enabled: true # Enables multinode orchestration via PodCliqueSets	CODE
MEDIUM	deploy/helm/charts/platform/values.yaml	54	# -- Whether to enable Grove integration (multinode orchestration via PodCliqueSets).	COMMENT
MEDIUM	deploy/helm/charts/platform/values.yaml	281	# Grove component - distributed inference orchestration	COMMENT
MEDIUM	deploy/operator/cmd/main.go	86	//+kubebuilder:scaffold:imports	COMMENT
MEDIUM	deploy/operator/cmd/main.go	165	//+kubebuilder:scaffold:scheme	COMMENT
MEDIUM⚡	deploy/operator/config/crd/kustomization.yaml	26	#+kubebuilder:scaffold:crdkustomizeresource	COMMENT
MEDIUM⚡	deploy/operator/config/crd/kustomization.yaml	31	#+kubebuilder:scaffold:crdkustomizewebhookpatch	COMMENT
MEDIUM⚡	deploy/operator/config/crd/kustomization.yaml	36	#+kubebuilder:scaffold:crdkustomizecainjectionpatch	COMMENT
MEDIUM	deploy/operator/config/samples/kustomization.yaml	24	#+kubebuilder:scaffold:manifestskustomizesamples	COMMENT
MEDIUM	deploy/operator/internal/controller/suite_test.go	56	//+kubebuilder:scaffold:imports	COMMENT
MEDIUM	deploy/operator/internal/controller/suite_test.go	104	//+kubebuilder:scaffold:scheme	COMMENT
MEDIUM	tests/frontend/test_vllm.py	99	"--max-model-len", # this reduced max context window and amount of GPU memory allocated for context	CODE
MEDIUM	tests/frontend/test_frontend_api_surface_compliance.py	551	# hallucination — it must actually invoke `ls` to discover it.	COMMENT
MEDIUM	tests/frontend/test_tool_calling_sglang.py	735	# hallucinate values for optional fields. Validate only that the	COMMENT
MEDIUM⚡	tests/serve/test_vllm.py	1095	# 1. Load-balances across N workers serving the SAME embedding model.	COMMENT
MEDIUM	tests/mm_router/test_mm_router_e2e.py	41	# Broad guardrails for TRT-LLM + Qwen3-VL-2B under block size 32.	COMMENT
MEDIUM	docs/design-docs/architecture.md	55	### Control Plane (adaptation and orchestration path)	COMMENT
MEDIUM	docs/backends/sglang/sglang-reference-guide.md	40	\| `--dyn-reasoning-parser` \| `DYN_REASONING_PARSER` \| `None` \| [Reasoning](../../reasoning/README.md#supported-reasoning	CODE
MEDIUM	…cs/backends/trtllm/trtllm-building-custom-container.md	87	If your custom build places TRT-LLM's bundled NIXL at a different path (or uses a non-3.12 Python), edit the `LD_PRELOAD	CODE
MEDIUM	docs/digest/agentic-inference/agentic-inference.md	133	The [NeMo Agent Toolkit (NAT)](https://github.com/NVIDIA/NeMo-Agent-Toolkit/tree/develop/examples/dynamo_integration) te	CODE
MEDIUM	docs/digest/tokenspeed/tokenspeed-day-0.md	11	[TokenSpeed](https://lightseek.org/blog/lightseek-tokenspeed.html) ([GitHub](https://github.com/lightseekorg/tokenspeed)	CODE
MEDIUM	docs/kubernetes/tilt-dev-setup.md	99	# enable_grove: false # PodClique-based multi-node orchestration	COMMENT
MEDIUM	docs/kubernetes/api-reference.md	971	\| `extraClientPods` _[GMSClientPodSpec](#gmsclientpodspec) array_ \| ExtraClientPods declares additional GMS client pods	CODE
MEDIUM	docs/kubernetes/api-reference.md	2518	\| `extraClientPods` _[GMSClientPodSpec](#gmsclientpodspec) array_ \| extraClientPods declares additional GMS client pods	CODE
MEDIUM	…ntainer/compliance/base_sboms/capture_baseline_sbom.py	540	# ---- main capture orchestration ----------------------------------------------	COMMENT
MEDIUM	components/src/dynamo/frontend/tests/FRONTEND_CASES.md	138	## `FRONTEND.9` — Reasoning ↔ tool-call orchestration	COMMENT
MEDIUM⚡	…rc/dynamo/frontend/tests/test_sglang_processor_unit.py	2717	class TestReasoningParsing: # FRONTEND.9 — reasoning ↔ tool-call orchestration	CODE
MEDIUM⚡	components/src/dynamo/vllm/llm_engine.py	692	# can push snapshots inline. Push is event-driven — no poll.	COMMENT
MEDIUM	components/src/dynamo/sglang/llm_engine.py	146	# it on every ZMQ message — event-driven, no framework polling.	COMMENT
MEDIUM	components/src/dynamo/replay/api.py	59	# offline planner run (matches the CLI's guardrails).	COMMENT
MEDIUM	fern/components/RecipeStyles.tsx	172	#workload-agentic-coding:checked ~ .dynamo-recipe-browser label[for="workload-agentic-coding"],	COMMENT
MEDIUM	fern/components/RecipeStyles.tsx	631	#workload-agentic-coding:checked ~ .dynamo-model-grid [data-recipe-card]:not([data-workload~="agentic-coding"]),	COMMENT
MEDIUM	recipes/nemotron-3-super/README.md	84	USECASE=chat # or agentic	CODE
MEDIUM	…pes/nemotron-3-super/vllm/agg-b200-agentic/deploy.yaml	5	# B200x4 (TP=4) tuned for the agentic-coding workload: DeepEP low-latency	COMMENT
MEDIUM	recipes/kimi-k2.6/README.md	76	USECASE=chat # or agentic	CODE
MEDIUM	recipes/kimi-k2.6/perf/README.md	43	The Dynamo Kimi-K2.5 recipe is the closest reference for trace handling — see [its README](https://github.com/ai-dynamo/	CODE
MEDIUM	recipes/kimi-k2.6/perf/README.md	43	The Dynamo Kimi-K2.5 recipe is the closest reference for trace handling — see [its README](https://github.com/ai-dynamo/	CODE
MEDIUM⚡	recipes/kimi-k2.5/tokenspeed/agg/nvidia/deploy.yaml	22	# the K8s-native discovery client requires operator-stamped scaffolding	COMMENT
MEDIUM⚡	recipes/kimi-k2.5/tokenspeed/agg/nvidia/deploy.yaml	25	# Replicating that scaffolding by hand defeats the "no operator required"	COMMENT
MEDIUM	recipes/nemotron-3-ultra/README.md	89	USECASE=chat # or agentic	CODE
MEDIUM	examples/backends/vllm/deploy/gaie/disagg.yaml	37	# * ``--max-model-len 16384`` caps the context window so vLLM's KV cache	COMMENT
MEDIUM	examples/backends/vllm/launch/agg_embed.sh	5	# Aggregated embedding model serving.	COMMENT
MEDIUM	examples/backends/vllm/launch/agg_embed.sh	15	# Default embedding model. Smaller alternatives:	COMMENT
MEDIUM	examples/backends/vllm/launch/disagg_omni_glm_image.sh	8	# Router: orchestrates the 2-stage pipeline, formats response	COMMENT
MEDIUM	examples/backends/vllm/launch/disagg_omni_glm_image.sh	78	# Router — discovers stage workers, orchestrates pipeline, formats response	COMMENT
MEDIUM	examples/backends/sglang/launch/agg_embed.sh	5	# Aggregated embedding model serving.	COMMENT
MEDIUM	.github/workflows/shared-compliance-audit.yml	49	# (sized to orchestrate REMOTE BuildKit, not scan locally)	COMMENT
MEDIUM	lib/bench/tests/request_trace.rs	4	//! End-to-end direct replay regressions for realistic agentic and multi-shard traces.	COMMENT
MEDIUM	lib/llm/tests/test_streaming_tool_parsers.rs	1392	// These reproduce the customer-reported issue (DIS-1765): Kimi K2 agentic	COMMENT
MEDIUM	lib/llm/tests/postprocessor_parsing_stream.rs	844	/// on the first chunk, so without guardrails the constrained JSON tool-call	COMMENT
MEDIUM⚡	…tests/data/deepseek-v3.2/test_input_search_w_date.json	441	"content": "[15] Zero-shot IR benchmark (Search_Results/Zero-shot IR benchmark)\n**viewing lines [0 - 40] of	CODE
MEDIUM⚡	…tests/data/deepseek-v3.2/test_input_search_w_date.json	441	"content": "[15] Zero-shot IR benchmark (Search_Results/Zero-shot IR benchmark)\n**viewing lines [0 - 40] of	CODE
MEDIUM⚡	…tests/data/deepseek-v3.2/test_input_search_w_date.json	446	"content": "[16] WebGPT benchmark (Search_Results/WebGPT benchmark)\nviewing lines [0 - 38] of 38\n\nL0:	CODE
MEDIUM⚡	…tests/data/deepseek-v3.2/test_input_search_w_date.json	456	"content": "[18] ALFWorld benchmark (Search_Results/ALFWorld benchmark)\nviewing lines [0 - 31] of 31\n\	CODE
MEDIUM⚡	…tests/data/deepseek-v3.2/test_input_search_w_date.json	642	"content": "[25] (https://ar5iv.org/abs/2307.13854)\nviewing lines [0 - 124] of 1092\n\nL0: \nL1: URL:	CODE
MEDIUM⚡	…tests/data/deepseek-v3.2/test_input_search_w_date.json	647	"content": "[26] (https://ar5iv.org/abs/2311.12983)\nviewing lines [0 - 118] of 1207\n\nL0: \nL1: URL:	CODE
MEDIUM⚡	…tests/data/deepseek-v3.2/test_input_search_w_date.json	647	"content": "[26] (https://ar5iv.org/abs/2311.12983)\nviewing lines [0 - 118] of 1207\n\nL0: \nL1: URL:	CODE
MEDIUM	lib/llm/src/http/service/anthropic.rs	346	// For agentic flows the model needs to see why it made prior decisions.	COMMENT
MEDIUM	lib/llm/src/protocols/common/extensions.rs	71	/// Identity metadata for agentic workloads.	COMMENT
7 more matches not shown…

Redundant / Tautological Comments119 hits · 178 pts

Severity	File	Line	Snippet	Context
LOW	deploy/utils/dynamo_deployment.py	477	# Check if deployment is ready	COMMENT
LOW	deploy/utils/kubernetes.py	65	# Check if pod exists and delete it if it does	COMMENT
LOW	deploy/utils/kubernetes.py	114	# Check if pod already exists and is running	COMMENT
LOW	deploy/utils/setup_benchmarking_resources.sh	49	# Check if namespace exists	COMMENT
LOW	deploy/utils/setup_benchmarking_resources.sh	55	# Check if Dynamo platform is installed	COMMENT
LOW	deploy/pre-deployment/pre-deployment-check.sh	111	# Check if there are multiple default storage classes (which can cause issues)	COMMENT
LOW	deploy/pre-deployment/pre-deployment-check.sh	146	# Check if GPU operator pods exist and are running	COMMENT
LOW	deploy/pre-deployment/pre-deployment-check.sh	157	# Check if any GPU operator pods are running	COMMENT
LOW	deploy/pre-deployment/nixl/build_and_deploy.sh	257	# Check if at least one valid step was selected	COMMENT
LOW	deploy/pre-deployment/nixl/build_and_deploy.sh	387	# Check if deployment file exists	COMMENT
LOW	…/controller/dynamographdeploymentrequest_controller.go	211	# Check if profiler container terminated	COMMENT
LOW	…ploy/operator/api/scripts/generate_pydantic_from_go.py	438	# Check if it's a known enum	COMMENT
LOW	…ploy/operator/api/scripts/generate_pydantic_from_go.py	445	# Check if it's a struct we're defining	COMMENT
LOW	tests/conftest.py	1009	# Check if a process is already running on the stored port	COMMENT
LOW	tests/conftest.py	895	asyncio.get_running_loop() # Check if we're in async context	CODE
LOW	tests/utils/managed_deployment.py	1535	# Check if port is assigned	COMMENT
LOW	tests/utils/managed_deployment.py	721	) # Check if process exists	CODE
LOW⚡	tests/utils/managed_process.py	580	# Check if any signaled group is still alive	COMMENT
LOW	tests/utils/managed_process.py	665	# Check if the main process is still alive	COMMENT
LOW	tests/utils/payloads.py	92	# Check if content is empty	COMMENT
LOW	tests/utils/payloads.py	99	# Check if ANY of the expected strings are found (OR logic) and count matches	COMMENT
LOW⚡	tests/fault_tolerance/cancellation/utils.py	127	# Set socket to non-blocking to avoid hanging	COMMENT
LOW	…_service/agents/gpu_fault_injector/gpu_xid_injector.py	181	# Check if proc path exists	COMMENT
LOW⚡	…e/hardware/fault_injection_service/api_service/main.py	1006	# Check if it's a NetworkPolicy	COMMENT
LOW⚡	…e/hardware/fault_injection_service/api_service/main.py	1009	# Check if it's a ChaosMesh resource	COMMENT
LOW	…ction_service/cuda_fault_injection/inject_into_pods.py	798	# Check if we have the expected number of pods	COMMENT
LOW	…ction_service/cuda_fault_injection/inject_into_pods.py	800	# Check if all pods are running	COMMENT
LOW	…ction_service/cuda_fault_injection/inject_into_pods.py	870	# Check if library exists (try both /tmp and /cuda-fault paths)	COMMENT
LOW	…ware/fault_injection_service/helpers/k8s_operations.py	203	# Check if pod is ready	COMMENT
LOW	…ware/fault_injection_service/helpers/k8s_operations.py	365	# Check if ready (all containers must be ready)	COMMENT
LOW	…dware/fault_injection_service/helpers/gpu_discovery.py	115	# Check if our PID is in the list	COMMENT
LOW	tests/fault_tolerance/deploy/scenarios.py	1127	# Check if this is an MoE deployment	COMMENT
LOW	tests/fault_tolerance/deploy/parse_results.py	193	# Check if this is a token overflow test	COMMENT
LOW	tests/fault_tolerance/deploy/parse_results.py	261	# Check if this is a mixed token test	COMMENT
LOW	tests/fault_tolerance/deploy/parse_results.py	885	# Check if this is an overflow/recovery pair and show timing info	COMMENT
LOW	tests/fault_tolerance/deploy/legacy_parse_results.py	104	# Check if request was successful	COMMENT
LOW	tests/fault_tolerance/deploy/test_deployment.py	151	# Check if this is a continuous load test (rolling upgrade scenarios)	COMMENT
LOW	tests/fault_tolerance/deploy/test_deployment.py	154	# Check if this is a mixed token test (overflow + recovery)	COMMENT
LOW	tests/fault_tolerance/etcd_ha/utils.py	226	# Check if a leader is elected indicating cluster health	COMMENT
LOW	tests/fault_tolerance/migration/utils.py	225	# Check if the worker logs contain the pattern	COMMENT
LOW	tests/kvbm_integration/common.py	650	# Check if process is listening on our port	COMMENT
LOW	tests/kvbm_integration/test_consolidator_router_e2e.py	30	# Check if engines are available and build list of available engines	COMMENT
LOW	tests/kvbm_integration/test_consolidator_router_e2e.py	206	# Check if consolidator started	COMMENT
LOW	tests/serve/lora_utils.py	144	# Check if MinIO is already running	COMMENT
LOW	tests/serve/launch/template_verifier.py	30	# Check if the custom template marker is present	COMMENT
LOW	tests/lmcache/summarize_scores_dynamo.py	185	# Check if this model passed consistency check	COMMENT
LOW	tests/lmcache/run_test.sh	30	# Check if data exists	COMMENT
LOW	tests/lmcache/run_test.sh	34	# Check if Python dependencies are installed	COMMENT
LOW	tests/lmcache/run_test.sh	63	# Check if server is responding	COMMENT
LOW	tests/lmcache/run_test.sh	104	# Check if server is responding	COMMENT
LOW	tests/lmcache/run_test.sh	141	# Check if result files exist	COMMENT
LOW	tests/router/helper.py	780	# Check if response has token_ids	COMMENT
LOW	tests/router/helper.py	810	# Verify if expected number of tokens are generated if max_tokens specified and ignore_eos is True	COMMENT
LOW	container/compliance/generate_go_deps.py	267	# Check if the package path starts with the module path	COMMENT
LOW	components/src/dynamo/frontend/vllm_processor.py	219	# Set DYNAMO_MM_TRANSFER to choose transfer mode:	COMMENT
LOW	components/src/dynamo/planner/core/load/predictors.py	154	# Check if all values are the same (constant data)	COMMENT
LOW	components/src/dynamo/planner/plugins/_proto_bridge.py	269	# Check if THIS message has a oneof named "result"	COMMENT
LOW	…onents/src/dynamo/planner/tests/unit/load_generator.py	381	# Print results summary	COMMENT
LOW	components/src/dynamo/vllm/worker_factory.py	903	# Check if kv event consolidator is enabled (port was allocated in setup_vllm_engine)	COMMENT
LOW	components/src/dynamo/vllm/worker_factory.py	1145	# Check if kv event consolidator is enabled (port was allocated in setup_vllm_engine)	COMMENT
59 more matches not shown…

Self-Referential Comments57 hits · 170 pts

Severity	File	Line	Snippet	Context
MEDIUM	hatch_build.py	59	version_content = f'# SPDX-FileCopyrightText: Copyright (c) 2025-2026 NVIDIA CORPORATION & AFFILIATES. All righ	CODE
MEDIUM	…rence-gateway/scripts/install_gaie_crd_agentgateway.sh	49	# Create an AgentgatewayParameters resource that excludes Istio sidecar injection	COMMENT
MEDIUM	deploy/helm/charts/snapshot/values.yaml	53	# Create a new PVC (set to false if using existing PVC)	COMMENT
MEDIUM	deploy/operator/config/crd/kustomizeconfig.yaml	16	# This file is for teaching kustomize how to substitute name and namespace reference in CRD	COMMENT
MEDIUM	deploy/operator/samples/dgd-gms-failover.yaml	13	# This file contains two commented-out variants you can copy into .spec.services:	COMMENT
MEDIUM	deploy/operator/samples/v1beta1/dgd-gms-failover.yaml	13	# This file contains variants you can use under .spec.components:	COMMENT
MEDIUM	tests/conftest.py	1042	# Create a log directory for session-scoped servers	COMMENT
MEDIUM	tests/conftest.py	1065	# Create a log directory for session-scoped servers	COMMENT
MEDIUM	tests/frontend/grpc/triton_echo_client.py	91	# Define the callback function. Note the last two parameters should be	COMMENT
MEDIUM	tests/frontend/grpc/test_tensor_mocker_engine.py	194	# Define the callback function. Note the last two parameters should be	COMMENT
MEDIUM	tests/utils/managed_deployment.py	1571	# Create a fresh portforward object so local_port=0 picks a new	COMMENT
MEDIUM	…e/hardware/fault_injection_service/api_service/main.py	769	# Create the NetworkPolicy	COMMENT
MEDIUM	…e/hardware/fault_injection_service/api_service/main.py	966	# Create the NetworkChaos resource	COMMENT
MEDIUM	tests/fault_tolerance/deploy/scenarios.py	819	# Define the yaml files for agg and disagg deployments	COMMENT
MEDIUM	tests/fault_tolerance/deploy/scenarios.py	825	# Define the different configurations to test	COMMENT
MEDIUM	tests/fault_tolerance/deploy/test_deployment.py	100	# Create a copy of the load config with overridden client type	COMMENT
MEDIUM	tests/lmcache/download_mmlu.py	68	# Create the expected CSV format: question, A, B, C, D, answer	COMMENT
MEDIUM	.devcontainer/gen_devcontainer_json.py	17	# Define the frameworks to generate	COMMENT
MEDIUM	.devcontainer/gen_devcontainer_json.py	29	# Create the target directory	COMMENT
MEDIUM	container/context.yaml	4	# This file represents the default ARG values of Dockerfiles generated	COMMENT
MEDIUM	components/src/dynamo/vllm/publisher.py	17	# Create a dedicated registry for dynamo_component metrics	COMMENT
MEDIUM⚡	…ests/memory/test_multimodal_embedding_cache_manager.py	262	# Create a non-contiguous tensor via transpose	COMMENT
MEDIUM	…amo/common/tests/multimodal/test_mm_kwargs_transfer.py	317	# Create a receiver with a mocked pool	COMMENT
MEDIUM⚡	…ponents/src/dynamo/common/config_dump/config_dumper.py	243	# Create a canonical JSON encoder with consistent formatting	COMMENT
MEDIUM	…nts/src/dynamo/common/multimodal/embedding_transfer.py	426	# Create a queue hinting whether the sender is expecting future transfer	COMMENT
MEDIUM	components/src/dynamo/mocker/main.py	163	# Create a separate DistributedRuntime for this worker (on same event loop)	COMMENT
MEDIUM	components/src/dynamo/mocker/main.py	215	# Create the engine with this worker's isolated runtime	COMMENT
MEDIUM	components/src/dynamo/sglang/args.py	526	# Create a minimal ServerArgs-like object that bypasses model config loading	COMMENT
MEDIUM	…mo/sglang/tests/test_sglang_image_diffusion_handler.py	131	# Create a simple test image	COMMENT
MEDIUM	…mo/sglang/tests/test_sglang_image_diffusion_handler.py	172	# Create a simple test image	COMMENT
MEDIUM	…mo/sglang/tests/test_sglang_image_diffusion_handler.py	284	# Create a numpy array representing an image	COMMENT
MEDIUM	components/src/dynamo/profiler/utils/plot.py	242	# Create the surface plot with customizations	COMMENT
MEDIUM	components/src/dynamo/profiler/utils/plot.py	278	# Create the throughput surface plot with customizations	COMMENT
MEDIUM	components/src/dynamo/trtllm/publisher.py	44	# Create a dedicated registry for dynamo_component metrics	COMMENT
MEDIUM	…ts/src/dynamo/trtllm/workers/image_diffusion_worker.py	75	# Create the request handler	COMMENT
MEDIUM	…ts/src/dynamo/trtllm/workers/video_diffusion_worker.py	74	# Create the request handler	COMMENT
MEDIUM	examples/backends/vllm/deploy/gms-failover.yaml	13	# This file contains two commented-out variants you can copy into .spec.services:	COMMENT
MEDIUM	examples/backends/vllm/deploy/v1beta1/gms-failover.yaml	13	# This file contains variants you can use under .spec.components:	COMMENT
MEDIUM	benchmarks/prefix_data_generator/cli.py	29	# Create the parser for the "analyze" command	COMMENT
MEDIUM	benchmarks/prefix_data_generator/cli.py	32	# Create the parser for the "synthesize" command	COMMENT
MEDIUM	…hmarks/prefix_data_generator/tests/test_synthesizer.py	76	# Create a temporary JSONL file with the specified data	COMMENT
MEDIUM	…hmarks/prefix_data_generator/tests/test_synthesizer.py	85	# Create the Synthesizer with the temporary file	COMMENT
MEDIUM⚡	benchmarks/prefix_data_generator/tests/test_sampler.py	23	# Create a test array with equal numbers of 1, 2, and 3	COMMENT
MEDIUM⚡	benchmarks/prefix_data_generator/tests/test_sampler.py	26	# Create the sampler	COMMENT
MEDIUM	benchmarks/frontend/scripts/sweep_k8s/aiperf.py	300	# Create a helper pod to access the PVC	COMMENT
MEDIUM	benchmarks/frontend/scripts/sweep_k8s/aiperf.py	460	# Create the Job	STRING
MEDIUM	benchmarks/multimodal/image/aiperf_large_image.sh	10	# Create a JSONL file with 11 identical large image URLs	COMMENT
MEDIUM	…ks/multimodal/image/aiperf_small_image_50_isl_ratio.sh	12	# Create a JSONL file with 30 identical small image URLs	COMMENT
MEDIUM	benchmarks/multimodal/image/aiperf_small_image.sh	11	# Create a JSONL file with 12 identical small image URLs	COMMENT
MEDIUM	.github/codeowners/external_contributors.yaml	14	# This file is also the single source for CONTRIBUTORS.md. Do not hand-edit	COMMENT
MEDIUM	.github/workflows/detect_broken_links.py	937	# Create a cleaned version of the report without resolved_path for JSON output	STRING
MEDIUM	.github/workflows/detect_broken_links.py	942	# Create a copy of the link without resolved_path	STRING
MEDIUM	.github/actions/docker-remote-build/action.yml	242	# Create the job metrics file	COMMENT
MEDIUM	.github/actions/docker-build/action.yml	227	# Create the job metrics file	COMMENT
MEDIUM	…gpu_memory_service/integrations/sglang/model_loader.py	38	# This module is only imported when load_format="gms" is used.	COMMENT
MEDIUM	lib/bindings/python/tests/test_kv_bindings.py	47	# Create a store event with parent_hash=None, block_hash=0	COMMENT
MEDIUM	lib/bindings/python/src/dynamo/nixl_connect/__init__.py	686	# This method is a no-op for now, in the future it may be used to initialize the connection.	COMMENT

Hallucination Indicators13 hits · 145 pts

Severity	File	Line	Snippet	Context
CRITICAL⚡	deploy/power-agent/tests/test_apply_cap.py	138	self.metrics.cap_clamped_total.labels.return_value.inc.assert_called_once()	CODE
CRITICAL⚡	deploy/power-agent/tests/test_apply_cap.py	146	self.metrics.cap_clamped_total.labels.return_value.inc.assert_called_once()	CODE
CRITICAL⚡	deploy/power-agent/tests/test_apply_cap.py	215	metrics.applied_limit_watts.labels.return_value.set.assert_called_once_with(450)	CODE
CRITICAL	deploy/power-agent/tests/test_actuator_protocol.py	353	metrics.applied_limit_watts.labels.return_value.set.assert_called_with(410)	CODE
CRITICAL	deploy/power-agent/tests/test_actuator_protocol.py	466	metrics.applied_limit_watts.labels.return_value.set.assert_called_with(700)	CODE
CRITICAL	deploy/power-agent/tests/test_dcgm_actuator.py	299	handle.GetSystem.return_value.discovery.GetAllGpuIds.assert_called_once_with()	CODE
CRITICAL	deploy/power-agent/tests/test_dcgm_actuator.py	1214	metrics.applied_limit_watts.labels.return_value.set.assert_called_with(300)	CODE
CRITICAL	deploy/power-agent/tests/test_dcgm_actuator.py	2585	metrics.applied_limit_watts.labels.return_value.set.assert_called_with(410)	CODE
CRITICAL	deploy/power-agent/tests/test_dcgm_actuator.py	2606	metrics.applied_limit_watts.labels.return_value.set.assert_called_with(700)	CODE
CRITICAL	…onents/src/dynamo/sglang/tests/test_sglang_rl_mixin.py	218	self.handler.engine.tokenizer_manager.flush_cache.assert_awaited_once_with()	CODE
CRITICAL	…onents/src/dynamo/sglang/tests/test_sglang_rl_mixin.py	232	self.handler.engine.tokenizer_manager.some_method.assert_awaited_once_with(	CODE
CRITICAL	…onents/src/dynamo/sglang/tests/test_sglang_rl_mixin.py	252	self.handler.engine.tokenizer_manager.some_method.assert_awaited_once_with(	CODE
CRITICAL	…onents/src/dynamo/sglang/tests/test_sglang_rl_mixin.py	291	self.handler.engine.tokenizer_manager.init_weights_update_group.assert_awaited_once_with(	CODE

Example Usage Blocks47 hits · 72 pts

Severity	File	Line	Snippet	Context
LOW	…lates/vllm/run_bare_multinode_elastic_ep_scale_test.sh	24	# Usage:	COMMENT
LOW	…nce/deploy/templates/vllm/run_elastic_ep_scale_test.sh	16	# Usage:	COMMENT
LOW	…/templates/vllm/run_multinode_elastic_ep_scale_test.sh	28	# Usage:	COMMENT
LOW	docs/digest/flash-indexer/tools/build.sh	11	# Usage:	COMMENT
LOW	fern/release_dryrun.sh	22	# Usage:	COMMENT
LOW	recipes/qwen3-vl-32b-fp8/benchmark/run-benchmark.sh	9	# Usage:	COMMENT
LOW	…qwen3-vl-30b/vllm/agg-embedding-cache/run-benchmark.sh	5	# Usage:	COMMENT
LOW	recipes/qwen3.6-35b/run-all-benchmarks.sh	12	# Usage:	COMMENT
LOW	recipes/qwen3.6-35b/run-benchmark.sh	12	# Usage:	COMMENT
LOW	examples/custom_encoder/launch/agg_custom.sh	26	# Usage:	COMMENT
LOW	examples/backends/vllm/launch/agg_omni_i2v.sh	7	# Usage:	COMMENT
LOW	examples/backends/vllm/launch/multi_node_tp.sh	9	# Usage:	COMMENT
LOW	examples/backends/vllm/launch/agg_embed_multiworker.sh	19	# Usage:	COMMENT
LOW	…kends/vllm/launch/lora/multimodal/validate_lora_agg.sh	13	# Usage:	COMMENT
LOW⚡	…mples/backends/vllm/launch/lora/multimodal/lora_agg.sh	11	# Usage:	COMMENT
LOW	examples/common/gpu_utils.sh	7	# Usage:	COMMENT
LOW	examples/common/gpu_utils.sh	22	# Usage:	COMMENT
LOW	examples/common/gpu_utils.sh	88	# Usage:	COMMENT
LOW	examples/common/launch_utils.sh	19	# Usage:	COMMENT
LOW⚡	examples/common/launch_utils.sh	63	# Usage:	COMMENT
LOW	examples/common/launch_utils.sh	283	# Usage:	COMMENT
LOW	examples/common/launch_utils.sh	313	# Usage:	COMMENT
LOW	examples/global_planner/global-planner-gpu-budget.yaml	27	# Usage:	COMMENT
LOW	examples/global_planner/global-planner-vllm-test.yaml	19	# Usage:	COMMENT
LOW	examples/global_planner/global-planner-mocker-test.yaml	15	# Usage:	COMMENT
LOW	…s/global_planner/global-planner-vllm-test-xpu-dra.yaml	21	# Usage:	COMMENT
LOW	…/global_planner/v1beta1/global-planner-gpu-budget.yaml	27	# Usage:	COMMENT
LOW	…s/global_planner/v1beta1/global-planner-vllm-test.yaml	19	# Usage:	COMMENT
LOW	…global_planner/v1beta1/global-planner-mocker-test.yaml	15	# Usage:	COMMENT
LOW	benchmarks/frontend/dgd/templates/mocker.yaml	19	# Usage:	COMMENT
LOW	benchmarks/frontend/dgd/templates/vllm-gpt-oss-20b.yaml	32	# Usage:	COMMENT
LOW	benchmarks/frontend/dgd/templates/vllm.yaml	17	# Usage:	COMMENT
LOW	benchmarks/frontend/scripts/run_perf.sh	9	# Usage:	COMMENT
LOW	benchmarks/frontend/scripts/nsight/nsys_profile.sh	13	# Usage:	COMMENT
LOW	…marks/frontend/scripts/flamegraph/offcpu_flamegraph.sh	8	# Usage:	COMMENT
LOW	…chmarks/frontend/scripts/flamegraph/diff_flamegraph.sh	8	# Usage:	COMMENT
LOW	…nchmarks/frontend/scripts/flamegraph/cpu_flamegraph.sh	8	# Usage:	COMMENT
LOW	benchmarks/frontend/scripts/bpf/setup.sh	10	# Usage:	COMMENT
LOW	benchmarks/frontend/scripts/bpf/run.sh	7	# Usage:	COMMENT
LOW	…modal/sweep/experiments/embedding_cache/trtllm_e_pd.sh	9	# Usage:	COMMENT
LOW	…odal/sweep/experiments/embedding_cache/vllm_serve.yaml	6	# Usage:	COMMENT
LOW	…dal/sweep/experiments/embedding_cache/trtllm_e_pd.yaml	6	# Usage:	COMMENT
LOW	…modal/sweep/experiments/embedding_cache/vllm_e_pd.yaml	6	# Usage:	COMMENT
LOW	…dal/sweep/experiments/embedding_cache/sglang_e_pd.yaml	6	# Usage:	COMMENT
LOW	benchmarks/omni/image/aiperf_image_gen.sh	8	# Usage:	COMMENT
LOW	lib/llm/tests/kv_manager.rs	249	// // Example usage:	COMMENT
LOW	dev/docker-observability.yml	7	# Usage:	COMMENT

Cross-Language Confusion14 hits · 70 pts

Severity	File	Line	Snippet	Context
HIGH	deploy/utils/convert_api_version.py	309	misread as bool/null, so converted manifests round-trip through kubectl."""	STRING
HIGH	…ault_injection_service/helpers/cuda_fault_injection.py	436	f'mkdir -p /host-fault && echo "{toggle_value}" > /host-fault/cuda_fault_enabled && cat /host-fault/	CODE
HIGH	…ault_injection_service/helpers/cuda_fault_injection.py	543	'rm -f /host-fault/cuda_fault_enabled 2>/dev/null; echo "ok"',	CODE
HIGH	tests/kvbm_integration/test_determinism_disagg.py	323	f"pip install hf_transfer && HF_HUB_ENABLE_HF_TRANSFER=1 hf download {model}",	CODE
HIGH	…ynamo-interconnect-check/scripts/check_interconnect.py	285	"command -v nixlbench nixl_test 2>/dev/null; ls /usr/local/nixl 2>/dev/null",	CODE
HIGH	docs/recipes/_catalog/validate.py	25	mappings, block lists of scalars and of mappings, quoted/plain scalars, null,	STRING
HIGH	components/src/dynamo/planner/plugins/scheduler.py	191	# call (last_call_at == -inf) lag is undefined; pin at	COMMENT
HIGH	…src/dynamo/common/configuration/groups/runtime_args.py	171	help="[Deprecated for vLLM] Use --kv-transfer-config instead. For TRT-LLM, options: nixl, lmcache, kvbm, nul	CODE
HIGH	…onents/src/dynamo/profiler/utils/dgdr_v1beta1_types.py	208	description="Planner contains the raw Planner configuration passed to the Planner service. Its schema is defined	CODE
HIGH	…/src/dynamo/trtllm/tests/test_trtllm_logits_runtime.py	401	"""When `stream_ptr` is non-null, the adapter wraps the processor	STRING
HIGH	benchmarks/frontend/scripts/sweep_k8s/aiperf.py	63	apt-get update -qq && apt-get install -y -qq curl jq git procps 2>/dev/null	CODE
HIGH	benchmarks/frontend/scripts/sweep_k8s/aiperf.py	311	command: ["sh", "-c", "echo ready && sleep 300"]	CODE
HIGH⚡	dev/sanity_check.py	3433	print(" Runtime: (cd lib/bindings/python && maturin develop)")	CODE
HIGH⚡	dev/sanity_check.py	3447	" Runtime: (cd lib/bindings/python && maturin develop)",	CODE

AI Structural Patterns42 hits · 34 pts

Severity	File	Line	Context
LOW	tests/frontend/test_completion_mocker_engine.py	51	CODE
LOW	tests/frontend/test_vllm.py	178	CODE
LOW	tests/utils/managed_process.py	909	CODE
LOW	tests/utils/payloads.py	399	CODE
LOW	tests/utils/payload_builder.py	59	CODE
LOW	tests/utils/payload_builder.py	101	CODE
LOW	tests/utils/payload_builder.py	261	CODE
LOW	tests/utils/payload_builder.py	373	CODE
LOW	tests/fault_tolerance/deploy/scenarios.py	1011	CODE
LOW	tests/fault_tolerance/deploy/scenarios.py	1052	CODE
LOW	tests/router/test_router_e2e_with_sglang.py	63	CODE
LOW	tests/router/router_process.py	41	CODE
LOW	tests/router/test_router_e2e_with_trtllm.py	70	CODE
LOW	tests/router/test_router_e2e_with_vllm.py	88	CODE
LOW	tests/router/common.py	1820	CODE
LOW	tests/router/common.py	2798	CODE
LOW	tests/router/helper.py	700	CODE
LOW	tests/router/mocker_process.py	131	CODE
LOW	tests/router/mocker_process.py	537	CODE
LOW	…c/dynamo/planner/tests/unit/test_load_based_scaling.py	47	CODE
LOW	components/src/dynamo/vllm/handlers.py	973	CODE
LOW	components/src/dynamo/vllm/handlers.py	2808	CODE
LOW	components/src/dynamo/vllm/handlers.py	3267	CODE
LOW	…/src/dynamo/vllm/multimodal_utils/request_processor.py	213	CODE
LOW	components/src/dynamo/common/utils/prometheus.py	188	CODE
LOW	components/src/dynamo/mocker/__init__.py	43	CODE
LOW	components/src/dynamo/sglang/tests/test_sglang_unit.py	382	CODE
LOW	…ents/src/dynamo/profiler/tests/test_replay_optimize.py	220	CODE
LOW	…ents/src/dynamo/profiler/tests/test_replay_optimize.py	253	CODE
LOW	components/src/dynamo/profiler/utils/aiperf.py	75	CODE
LOW	…src/dynamo/profiler/utils/config_modifiers/protocol.py	128	CODE
LOW	…src/dynamo/profiler/utils/config_modifiers/protocol.py	473	CODE
LOW	components/src/dynamo/replay/api.py	27	CODE
LOW	components/src/dynamo/replay/api.py	130	CODE
LOW	components/src/dynamo/trtllm/llm_engine.py	126	CODE
LOW	…mponents/src/dynamo/trtllm/engines/diffusion_engine.py	158	CODE
LOW	fern/convert_callouts.py	124	CODE
LOW	benchmarks/frontend/scripts/sweep_k8s/aiperf.py	382	CODE
LOW	lib/gpu_memory_service/snapshot/storage_client.py	41	CODE
LOW	lib/bindings/python/src/dynamo/_internal/aic.py	367	CODE
LOW	lib/bindings/python/src/dynamo/_internal/aic.py	404	CODE
LOW	lib/bindings/python/src/dynamo/_internal/aic.py	172	CODE

Synthetic Comment Markers3 hits · 20 pts

Severity	File	Line	Snippet	Context
HIGH	docs/benchmarks/benchmarking.md	303	└── plots/ # Generated by aiperf plot	CODE
HIGH	lib/llm/src/kv_router/publisher/tests.rs	1728	#[ignore] // Mark as ignored as requested, because CI's integrations still don't have NATS	COMMENT
HIGH	lib/llm/src/kv_router/publisher/tests.rs	1728	#[ignore] // Mark as ignored as requested, because CI's integrations still don't have NATS	COMMENT

AI Response Leakage2 hits · 15 pts

Severity	File	Line	Snippet	Context
HIGH	examples/global_planner/global-planner-gpu-budget.yaml	12	# --managed-namespaces), not to every DGD in the cluster. In this example the	COMMENT
HIGH	…/global_planner/v1beta1/global-planner-gpu-budget.yaml	12	# --managed-namespaces), not to every DGD in the cluster. In this example the	COMMENT

Fake / Example Data12 hits · 14 pts

Severity	File	Line	Snippet	Context
LOW	deploy/operator/internal/webhook/common_test.go	116	username: "admin@example.com",	CODE
LOW	…/tests/multimodal_utils/test_vllm_request_processor.py	510	decode_mm_data = {"image": {"placeholder": object()}}	CODE
LOW⚡	.github/codeowners/test_codeowners.py	527	"name": "Jane Doe",	CODE
LOW⚡	.github/codeowners/test_codeowners.py	535	assert "Jane Doe" in md	CODE
LOW	lib/bench/multiturn_bench.rs	51	/// Approximate user-prompt token count per turn (lorem ipsum)	COMMENT
LOW	lib/bench/multiturn_bench.rs	137	// Lorem ipsum generator	COMMENT
LOW⚡	lib/llm/tests/tokenizers.rs	35	("Tell me about the following text.", "Lorem ipsum dolor sit amet, consectetur adipiscing elit, sed do eiusmod tempo	CODE
LOW⚡	lib/llm/tests/tokenizers.rs	35	("Tell me about the following text.", "Lorem ipsum dolor sit amet, consectetur adipiscing elit, sed do eiusmod tempo	CODE
LOW⚡	lib/llm/tests/tokenizers.rs	38	("Tell me about the following text.", "Neque porro quisquam est, qui dolorem ipsum quia dolor sit amet, consectetur,	CODE
LOW⚡	lib/llm/tests/tokenizers.rs	38	("Tell me about the following text.", "Neque porro quisquam est, qui dolorem ipsum quia dolor sit amet, consectetur,	CODE
LOW	…sts/data/replays/deepseek-r1-distill-llama-8b/notes.md	10	"messages": [{"role": "user", "content": "Lorem ipsum dolor sit amet, consectetur adipiscing elit, sed do eiusmod te	CODE
LOW	…sts/data/replays/deepseek-r1-distill-llama-8b/notes.md	10	"messages": [{"role": "user", "content": "Lorem ipsum dolor sit amet, consectetur adipiscing elit, sed do eiusmod te	CODE

Dead Code5 hits · 10 pts

Severity	File	Line	Context
MEDIUM	tests/frontend/http_status_propagation_worker.py	36	CODE
MEDIUM	components/src/dynamo/vllm/tests/test_vllm_lora.py	216	CODE
MEDIUM	…s/src/dynamo/vllm/tests/omni/test_omni_stage_worker.py	67	CODE
MEDIUM	components/src/dynamo/vllm/omni/base_handler.py	126	CODE
MEDIUM	…amo/trtllm/request_handlers/base_generative_handler.py	46	CODE

Slop Phrases3 hits · 7 pts

Severity	File	Line	Snippet	Context
MEDIUM	deploy/operator/samples/v1beta1/dgd-gms-failover.yaml	13	# This file contains variants you can use under .spec.components:	COMMENT
MEDIUM	examples/backends/vllm/deploy/v1beta1/gms-failover.yaml	13	# This file contains variants you can use under .spec.components:	COMMENT
LOW	lib/llm/tests/parallel_tool_call_integration.rs	103	r#"<think>Okay, the user is asking two things: the weather in Dallas, Texas, and whether today is a holiday. I need	CODE

TODO Padding1 hit · 2 pts

Severity	File	Line	Snippet	Context
LOW	lib/kvbm-engine/src/worker/coordinated.rs	392	// TODO: Add tests with mock Worker implementation	COMMENT

Overly Generic Function Names1 hit · 1 pts

Severity	File	Line	Snippet	Context
LOW	lib/gpu_memory_service/server/gms.py	229	async def handle_request(	CODE

Analysis Overview

What These Metrics Mean

Score History

Severity Breakdown

Directory Score Breakdown

Pattern Findings