antoinezambelli/forge

32.0

Adjusted Score

32.0

Raw Score

100%

Time Factor

2026-08-26

Last Push

2.2K

Stars

Python

Language

78.3K

Lines of Code

230

Files

1.9K

Pattern Hits

2026-08-29

Scan Date

0.07

HC Hit Rate

What These Metrics Mean

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

Score History

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

Severity Breakdown

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

CRITICAL 0HIGH 15MEDIUM 271LOW 1664

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 1950 distinct pattern matches across 17 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.

Hyper-Verbose Identifiers1249 hits · 1157 pts

Severity	File	Line	Snippet	Context
LOW	tests/conftest.py	17	def mock_httpx_client_constructor() -> Iterator[MagicMock]:	CODE
LOW	tests/http_mocks.py	32	def patch_httpx_client_constructor() -> Iterator[MagicMock]:	CODE
LOW	tests/unit/test_argument_transformation.py	39	def test_lambda_and_stateful_tools_share_output_contracts() -> None:	CODE
LOW	tests/unit/test_argument_transformation.py	180	def test_lambda_and_stateful_validators_accept_equivalent_formats() -> None:	CODE
LOW	tests/unit/test_argument_transformation.py	195	def test_lambda_and_stateful_validators_reject_benchmark_failure_modes() -> None:	CODE
LOW	tests/unit/test_argument_transformation.py	233	def test_stateful_tools_record_successful_reasoning_path() -> None:	CODE
LOW	tests/unit/test_argument_transformation.py	257	def test_stateful_tools_do_not_record_rejected_paths() -> None:	CODE
LOW	tests/unit/test_argument_transformation.py	267	def _exercise_required_stateful_calls(workflow: Any) -> None:	CODE
LOW	tests/unit/test_argument_transformation.py	278	def test_validate_state_requires_full_reasoning_path_and_submission() -> None:	CODE
LOW	tests/unit/test_argument_transformation.py	297	def test_validate_state_rejects_skipped_conversion() -> None:	CODE
LOW	tests/unit/test_argument_transformation.py	308	def test_validate_state_rejects_wrong_submission() -> None:	CODE
LOW⚡	tests/unit/test_openai_compat_client.py	90	async def test_returns_text_response(self) -> None:	CODE
LOW⚡	tests/unit/test_openai_compat_client.py	100	async def test_missing_choices_raises_backend_error(self) -> None:	CODE
LOW⚡	tests/unit/test_openai_compat_client.py	110	async def test_null_content_returns_empty_text(self) -> None:	CODE
LOW⚡	tests/unit/test_openai_compat_client.py	120	async def test_malformed_tool_args_kept_as_raw_args(self) -> None:	CODE
LOW⚡	tests/unit/test_openai_compat_client.py	454	def test_bearer_header_set_when_key_provided(self) -> None:	CODE
LOW⚡	tests/unit/test_openai_compat_client.py	460	def test_no_auth_header_when_no_key(self) -> None:	CODE
LOW⚡	tests/unit/test_openai_compat_client.py	464	def test_base_url_trailing_slash_stripped(self) -> None:	CODE
LOW⚡	tests/unit/test_openai_compat_client.py	474	async def test_yields_text_deltas_and_final(self) -> None:	CODE
LOW⚡	tests/unit/test_openai_compat_client.py	827	def test_recommended_sampling_off_is_silent_for_unknown_model(self) -> None:	CODE
LOW⚡	tests/unit/test_openai_compat_client.py	837	def test_recommended_sampling_on_raises_for_unknown_model(self) -> None:	CODE
LOW	tests/unit/test_openai_compat_client.py	142	async def test_malformed_tool_args_kept_even_with_assistant_text(self) -> None:	CODE
LOW	tests/unit/test_openai_compat_client.py	163	async def test_malformed_among_several_kept_per_call(self) -> None:	CODE
LOW	tests/unit/test_openai_compat_client.py	185	async def test_dict_tool_args_accepted(self) -> None:	CODE
LOW	tests/unit/test_openai_compat_client.py	204	async def test_formats_tools_in_request(self) -> None:	CODE
LOW	tests/unit/test_openai_compat_client.py	219	async def test_request_body_structure(self) -> None:	CODE
LOW	tests/unit/test_openai_compat_client.py	234	async def test_posts_to_chat_completions(self) -> None:	CODE
LOW	tests/unit/test_openai_compat_client.py	256	async def test_http_error_raises_backend_error(self) -> None:	CODE
LOW	tests/unit/test_openai_compat_client.py	265	async def test_structured_reasoning_field_captured(self, field) -> None:	CODE
LOW	tests/unit/test_openai_compat_client.py	286	async def test_extracts_think_tags_from_content_with_tool_call(self) -> None:	CODE
LOW	tests/unit/test_openai_compat_client.py	306	async def test_reasoning_field_preferred_over_content_tags(self) -> None:	CODE
LOW	tests/unit/test_openai_compat_client.py	325	async def test_content_preamble_not_treated_as_reasoning(self) -> None:	CODE
LOW	tests/unit/test_openai_compat_client.py	347	async def test_think_tags_stripped_from_text_response(self) -> None:	CODE
LOW	tests/unit/test_openai_compat_client.py	362	async def test_plain_preamble_text_response_survives(self) -> None:	CODE
LOW	tests/unit/test_openai_compat_client.py	377	async def test_reasoning_attached_to_first_tool_call_only(self) -> None:	CODE
LOW	tests/unit/test_openai_compat_client.py	398	async def test_empty_string_structured_field_falls_through_to_tags(self) -> None:	CODE
LOW	tests/unit/test_openai_compat_client.py	419	async def test_non_string_reasoning_field_raises(self) -> None:	CODE
LOW	tests/unit/test_openai_compat_client.py	496	async def test_yields_final_with_tool_call(self) -> None:	CODE
LOW	tests/unit/test_openai_compat_client.py	525	async def test_stream_malformed_tool_args_kept_as_raw_args(self) -> None:	CODE
LOW	tests/unit/test_openai_compat_client.py	552	async def test_stream_non_string_arg_fragment_not_dropped(self) -> None:	CODE
LOW	tests/unit/test_openai_compat_client.py	574	async def test_accumulates_reasoning_across_deltas(self) -> None:	CODE
LOW	tests/unit/test_openai_compat_client.py	597	async def test_stream_extracts_think_tags_from_content_with_tool_call(self) -> None:	CODE
LOW	tests/unit/test_openai_compat_client.py	620	async def test_stream_content_preamble_not_treated_as_reasoning(self) -> None:	CODE
LOW	tests/unit/test_openai_compat_client.py	643	async def test_stream_strips_think_tags_from_text_response(self) -> None:	CODE
LOW	tests/unit/test_openai_compat_client.py	661	async def test_stream_other_canonical_fields_accumulate(self, field) -> None:	CODE
LOW	tests/unit/test_openai_compat_client.py	682	async def test_stream_structured_preferred_over_content_tags(self) -> None:	CODE
LOW	tests/unit/test_openai_compat_client.py	703	async def test_stream_reasoning_first_tool_call_only(self) -> None:	CODE
LOW	tests/unit/test_openai_compat_client.py	725	async def test_stream_non_string_reasoning_raises(self) -> None:	CODE
LOW	tests/unit/test_openai_compat_client.py	741	async def test_stream_http_error_raises(self) -> None:	CODE
LOW	tests/unit/test_openai_compat_client.py	752	async def test_ignores_non_data_lines(self) -> None:	CODE
LOW	tests/unit/test_openai_compat_client.py	777	def test_extra_headers_merged_alongside_bearer(self) -> None:	CODE
LOW	tests/unit/test_openai_compat_client.py	789	def test_api_key_plus_auth_header_raises(self) -> None:	CODE
LOW	tests/unit/test_openai_compat_client.py	802	def test_extra_headers_alone_set_custom_scheme(self) -> None:	CODE
LOW	tests/unit/test_openai_compat_client.py	811	def test_sampling_kwargs_stored_as_instance_fields(self) -> None:	CODE
LOW	tests/unit/test_openai_compat_client.py	859	async def test_instance_sampling_flows_into_body(self, field, value) -> None:	CODE
LOW	tests/unit/test_openai_compat_client.py	881	async def test_passthrough_fields_appear_in_body(self) -> None:	CODE
LOW	tests/unit/test_openai_compat_client.py	896	async def test_forge_owned_fields_override_passthrough(self) -> None:	CODE
LOW	tests/unit/test_openai_compat_client.py	917	async def test_inbound_anthropic_body_accepted_and_ignored(self) -> None:	CODE
LOW	tests/unit/test_openai_compat_client.py	933	async def test_send_stream_accepts_passthrough_and_inbound(self) -> None:	CODE
LOW	tests/unit/test_openai_compat_client.py	954	async def test_aclose_closes_http_pool(self) -> None:	CODE
1189 more matches not shown…

Decorative Section Separators246 hits · 808 pts

Severity	File	Line	Snippet	Context
MEDIUM	tests/conftest.py	23	# ── Shared LLM client double ─────────────────────────────────────	COMMENT
MEDIUM⚡	tests/unit/test_openai_compat_client.py	450	# ── auth ─────────────────────────────────────────────────────────	COMMENT
MEDIUM⚡	tests/unit/test_openai_compat_client.py	469	# ── send_stream ──────────────────────────────────────────────────	COMMENT
MEDIUM⚡	tests/unit/test_openai_compat_client.py	847	# ── instance sampling flows into request body ────────────────────	COMMENT
MEDIUM	tests/unit/test_openai_compat_client.py	61	# ── send ─────────────────────────────────────────────────────────	COMMENT
MEDIUM	tests/unit/test_openai_compat_client.py	773	# ── constructor ──────────────────────────────────────────────────	COMMENT
MEDIUM	tests/unit/test_openai_compat_client.py	876	# ── passthrough + inbound_anthropic_body ─────────────────────────	COMMENT
MEDIUM	tests/unit/test_openai_compat_client.py	949	# ── aclose ───────────────────────────────────────────────────────	COMMENT
MEDIUM⚡	tests/unit/test_inconsistent_api_recovery.py	27	# ── Lambda tool callables ───────────────────────────────────────	COMMENT
MEDIUM⚡	tests/unit/test_inconsistent_api_recovery.py	243	# ── Stateful backend ────────────────────────────────────────────	COMMENT
MEDIUM⚡	tests/unit/test_inconsistent_api_recovery.py	327	# ── Stateful validate_state ─────────────────────────────────────	COMMENT
MEDIUM	tests/unit/test_inconsistent_api_recovery.py	200	# ── Validator (lambda) ──────────────────────────────────────────	COMMENT
MEDIUM⚡	tests/unit/test_ollama_client.py	849	# ── outbound message normalization (#111 / #115) ─────────────────	COMMENT
MEDIUM⚡	tests/unit/test_ollama_client.py	960	# ── get_context_length + set_num_ctx ─────────────────────────────	COMMENT
MEDIUM⚡	tests/unit/test_ollama_client.py	982	# ── format_tool ──────────────────────────────────────────────────	COMMENT
MEDIUM	tests/unit/test_ollama_client.py	47	# ── send ─────────────────────────────────────────────────────────	COMMENT
MEDIUM	tests/unit/test_ollama_client.py	476	# ── send_stream ──────────────────────────────────────────────────	COMMENT
MEDIUM⚡	tests/unit/test_proxy_auth.py	86	# ── relocate_credential ──────────────────────────────────────────────	COMMENT
MEDIUM⚡	tests/unit/test_proxy_auth.py	132	# ── resolve_inbound_credential ───────────────────────────────────────	COMMENT
MEDIUM	tests/unit/test_proxy_auth.py	27	# ── extract_inbound_credential ───────────────────────────────────────	COMMENT
MEDIUM	tests/unit/test_proxy_auth.py	224	# ── handler threading ────────────────────────────────────────────────	COMMENT
MEDIUM	tests/unit/test_anthropic_client.py	137	# ── _convert_tools ───────────────────────────────────────────────	COMMENT
MEDIUM	tests/unit/test_anthropic_client.py	177	# ── _convert_messages ────────────────────────────────────────────	COMMENT
MEDIUM	tests/unit/test_anthropic_client.py	450	# ── _parse_response ──────────────────────────────────────────────	COMMENT
MEDIUM	tests/unit/test_anthropic_client.py	727	# ── Prompt caching (static tools+system breakpoint) ──────────────	COMMENT
MEDIUM⚡	tests/unit/test_llamafile_client.py	1135	# ── _merge_consecutive ────────────────────────────────────────────	COMMENT
MEDIUM⚡	tests/unit/test_llamafile_client.py	1138	# ── think flag behavior — sync ────────────────────────────────────	COMMENT
MEDIUM	tests/unit/test_llamafile_client.py	160	# ── send — native mode ──────────────────────────────────────────	COMMENT
MEDIUM	tests/unit/test_llamafile_client.py	659	# ── send — prompt mode ──────────────────────────────────────────	COMMENT
MEDIUM	tests/unit/test_llamafile_client.py	749	# ── get_context_length ───────────────────────────────────────────	COMMENT
MEDIUM	tests/unit/test_llamafile_client.py	792	# ── send_stream ──────────────────────────────────────────────────	COMMENT
MEDIUM	tests/unit/test_llamafile_client.py	969	# ── send_stream — malformed-500 transport boundaries ──────────	COMMENT
MEDIUM	tests/unit/test_llamafile_client.py	1049	# ── mode ─────────────────────────────────────────────────────────	COMMENT
MEDIUM	tests/unit/test_llamafile_client.py	1059	# ── _apply_sampling ──────────────────────────────────────────────	COMMENT
MEDIUM	tests/unit/test_llamafile_client.py	1391	# ── think flag behavior — stream ──────────────────────────────────	COMMENT
MEDIUM	tests/unit/test_llamafile_client.py	1486	# ── slot_id ────────────────────────────────────────────────────	COMMENT
MEDIUM	tests/unit/test_llamafile_client.py	1658	# ── Issue #121: Path.stem truncates dotted model names ─────────────	COMMENT
MEDIUM⚡	tests/unit/test_guardrails.py	22	# ── check(): response validation ─────────────────────────────	COMMENT
MEDIUM⚡	tests/unit/test_guardrails.py	126	# ── check(): step enforcement ────────────────────────────────	COMMENT
MEDIUM⚡	tests/unit/test_guardrails.py	162	# ── record() ─────────────────────────────────────────────────	COMMENT
MEDIUM⚡	tests/unit/test_guardrails.py	191	# ── CheckResult dataclass ────────────────────────────────────	COMMENT
MEDIUM⚡	tests/unit/test_guardrails.py	220	# ── Custom retry nudge ───────────────────────────────────────	COMMENT
MEDIUM	tests/unit/test_guardrails.py	9	# ── Helpers ──────────────────────────────────────────────────	COMMENT
MEDIUM	tests/unit/test_guardrails.py	206	# ── Imports ──────────────────────────────────────────────────	COMMENT
MEDIUM⚡	tests/unit/test_proxy_path1.py	27	# ── ProxyServer construction validation ──────────────────────	COMMENT
MEDIUM	tests/unit/test_proxy_path1.py	94	# ── AnthropicClient verbatim path ────────────────────────────	COMMENT
MEDIUM	tests/unit/test_proxy_path1.py	182	# ── AnthropicClient base_url ─────────────────────────────────	COMMENT
MEDIUM	tests/unit/test_proxy_path1.py	198	# ── End-to-end: cache_control wire preservation ──────────────	COMMENT
MEDIUM⚡	tests/unit/test_proxy_convert_anthropic.py	19	# ── Inbound: Anthropic → forge ─────────────────────────────	COMMENT
MEDIUM	tests/unit/test_proxy_convert_anthropic.py	237	# ── Outbound: forge → Anthropic ────────────────────────────	COMMENT
MEDIUM	tests/unit/test_proxy_convert_anthropic.py	293	# ── SSE event sequence ─────────────────────────────────────	COMMENT
MEDIUM⚡	tests/unit/test_context_manager.py	89	# ── estimate_tokens ─────────────────────────────────────────────	COMMENT
MEDIUM⚡	tests/unit/test_context_manager.py	187	# ── maybe_compact — under threshold ─────────────────────────────	COMMENT
MEDIUM⚡	tests/unit/test_context_manager.py	200	# ── maybe_compact — over threshold ──────────────────────────────	COMMENT
MEDIUM	tests/unit/test_context_manager.py	14	# ── Helpers ──────────────────────────────────────────────────────	COMMENT
MEDIUM	tests/unit/test_context_manager.py	268	# ── CompactEvent fields ─────────────────────────────────────────	COMMENT
MEDIUM⚡	tests/unit/test_vllm_client.py	89	# ── constructor / identity derivation ──────────────────────────	COMMENT
MEDIUM⚡	tests/unit/test_vllm_client.py	119	# ── send (non-streaming) ──────────────────────────────────────	COMMENT
MEDIUM⚡	tests/unit/test_vllm_client.py	689	# ── get_served_model_name ──────────────────────────────────────	COMMENT
MEDIUM⚡	tests/unit/test_vllm_client.py	715	# ── removed proxy discovery helper ─────────────────────────────	COMMENT
186 more matches not shown…

Unused Imports247 hits · 212 pts

Severity	File	Line	Context
LOW	tests/conftest.py	3	CODE
LOW	tests/http_mocks.py	3	CODE
LOW	tests/unit/test_argument_transformation.py	3	CODE
LOW	tests/unit/test_inconsistent_api_recovery.py	13	CODE
LOW	tests/unit/test_proxy_installation_docs.py	3	CODE
LOW	tests/unit/test_eval_outcomes.py	3	CODE
LOW	tests/unit/test_standalone_build.py	3	CODE
LOW	tests/unit/test_eval_dataset_builder.py	3	CODE
LOW	tests/unit/test_clients_base.py	3	CODE
LOW	tests/unit/test_batch_eval_recipes.py	3	CODE
LOW	tests/unit/test_server_rpc.py	3	CODE
LOW	tests/unit/test_eval_reporting_metrics.py	3	CODE
LOW	tests/unit/test_proxy_profiles.py	3	CODE
LOW	tests/unit/test_proxy_path1.py	13	CODE
LOW	tests/unit/test_data_gap_recovery_extended.py	3	CODE
LOW	tests/unit/test_proxy_cli.py	3	CODE
LOW	tests/unit/test_proxy_installer.py	3	CODE
LOW	tests/unit/test_backend_profiles.py	3	CODE
LOW	tests/unit/test_server.py	3	CODE
LOW	tests/unit/test_proxy_config.py	3	CODE
LOW	tests/unit/test_slot_worker.py	3	CODE
LOW	tests/unit/test_slot_worker.py	6	CODE
LOW	tests/unit/test_slot_worker.py	8	CODE
LOW	tests/unit/test_slot_worker.py	14	CODE
LOW	tests/unit/test_slot_worker.py	389	CODE
LOW	tests/unit/test_runner.py	3	CODE
LOW	tests/unit/test_eval_budget.py	3	CODE
LOW	tests/unit/test_eval_budget.py	15	CODE
LOW	tests/unit/test_proxy_release.py	3	CODE
LOW	tests/unit/test_proxy_proxy.py	9	CODE
LOW	tests/unit/test_proxy_release_workflows.py	3	CODE
LOW	tests/unit/test_batch_eval_resume.py	3	CODE
LOW	tests/unit/test_rpc.py	3	CODE
LOW	tests/unit/test_client_auth.py	15	CODE
LOW	tests/unit/test_eval_publication.py	3	CODE
LOW	tests/unit/test_proxy_lifecycle_smoke.py	3	CODE
LOW	tests/unit/test_eval_generation.py	3	CODE
LOW	tests/unit/test_grounded_synthesis.py	12	CODE
LOW	tests/unit/test_batch_eval_lifecycle.py	3	CODE
LOW	tests/integration/_bootstrap_support.py	3	CODE
LOW	…egration/bootstrap_contract/test_bootstrap_contract.py	8	CODE
LOW	…egration/platform_acceptance/test_windows_installer.py	3	CODE
LOW	tests/eval/metrics.py	3	CODE
LOW	tests/eval/generation.py	7	CODE
LOW	tests/eval/batch_eval.py	11	CODE
LOW	tests/eval/batch_eval.py	19	CODE
LOW	tests/eval/batch_eval.py	33	CODE
LOW	tests/eval/batch_eval.py	38	CODE
LOW	tests/eval/ablation.py	3	CODE
LOW	tests/eval/dataset_builder.py	13	CODE
LOW	tests/eval/outcomes.py	8	CODE
LOW	tests/eval/significance.py	15	CODE
LOW	tests/eval/publication.py	15	CODE
LOW	tests/eval/eval_runner.py	3	CODE
LOW	tests/eval/eval_runner.py	12	CODE
LOW	tests/eval/eval_runner.py	18	CODE
LOW	tests/eval/provenance.py	3	CODE
LOW	tests/eval/report.py	17	CODE
LOW	tests/eval/scenarios/_base.py	3	CODE
LOW	tests/eval/scenarios/_stateful_plumbing.py	3	CODE
187 more matches not shown…

Modern AI Meta-Vocabulary23 hits · 63 pts

Severity	File	Line	Snippet	Context
MEDIUM	README.md	3	[![PyPI](https://img.shields.io/pypi/v/forge-guardrails.svg)](https://pypi.org/project/forge-guardrails/)	CODE
MEDIUM	README.md	15	- Not a coding harness. Forge is domain-agnostic. If you're building a coding agent (or already using one like openc	CODE
MEDIUM	README.md	174	Drop-in proxy that sits between any client and a local model server, speaking both the OpenAI chat-completions API and t	CODE
MEDIUM	README.md	325	runner.py # WorkflowRunner — the agentic loop	CODE
MEDIUM	CONTRIBUTING.md	122	### Adding or modifying guardrails	COMMENT
MEDIUM	tests/unit/test_openai_compat_client.py	421	# it into replayable chain-of-thought.	COMMENT
MEDIUM	tests/unit/test_openai_compat_client.py	727	# never repr-coerced into chain-of-thought.	COMMENT
MEDIUM	tests/eval/report.py	1327	("## Full Ablation — which guardrails do the work?", lambda rp: rp == "ablation.md"),	CODE
MEDIUM	tests/eval/scenarios/_model_reasoning.py	569	# guardrails-vs-bare ablation: bare runs hit the first error and	COMMENT
MEDIUM	docs/USER_GUIDE.md	138	### Mode 3: Middleware (composable guardrails)	COMMENT
MEDIUM	docs/WORKFLOW.md	474	└── reasoning: str \| None # chain-of-thought (thinking models)	CODE
MEDIUM	docs/EVAL_GUIDE.md	87	# Ablation — bare (all guardrails off)	COMMENT
MEDIUM	docs/decisions/MULTI_MODEL_ROUTING.md	170	# Consumer orchestration code	COMMENT
MEDIUM	src/forge/clients/sampling_defaults.py	80	# documented default reasoning effort for complex coding and agentic work.	COMMENT
MEDIUM	src/forge/clients/sampling_defaults.py	120	# DeepSeek-V4-Flash-0731 — agentic profile: T=1.0, top_p=0.95. The card	COMMENT
MEDIUM	src/forge/clients/sampling_defaults.py	190	# explicit chain of thought). Card recommends temperature=0.2, top_k=80, repetition_penalty=1.05	COMMENT
MEDIUM	src/forge/clients/openai_compat.py	397	# chain-of-thought.	COMMENT
MEDIUM	src/forge/proxy/handler.py	429	# in tool-calling mode where guardrails apply. The respond call is	COMMENT
MEDIUM	src/forge/proxy/handler.py	443	# No tools → plain chat completion, no guardrails needed.	COMMENT
MEDIUM	src/forge/proxy/handler.py	474	# Set up guardrails	COMMENT
MEDIUM	src/forge/proxy/handler.py	498	# error. The client's own agentic loop can decide what to do.	COMMENT
MEDIUM	src/forge/proxy/proxy.py	346	# guardrails wrap its inference loop like any other client.	COMMENT
MEDIUM	src/forge/proxy/proxy.py	348	# (forge-guardrails[anthropic]). Only Path 1 needs it, so	COMMENT

Deep Nesting43 hits · 41 pts

Severity	File	Line	Context
LOW	tests/unit/test_eval_dataset_builder.py	314	CODE
LOW	tests/unit/test_proxy_release.py	69	CODE
LOW	tests/eval/metrics.py	275	CODE
LOW	tests/eval/batch_eval.py	639	CODE
LOW	tests/eval/batch_eval.py	1065	CODE
LOW	tests/eval/dataset_builder.py	684	CODE
LOW	tests/eval/dataset_builder.py	746	CODE
LOW	tests/eval/significance.py	106	CODE
LOW	tests/eval/publication.py	465	CODE
LOW	tests/eval/eval_runner.py	413	CODE
LOW	tests/eval/report.py	278	CODE
LOW	tests/eval/report.py	762	CODE
LOW	scripts/run_ablation.py	102	CODE
LOW	scripts/migrate_eval_jsonl_gguf_identity.py	100	CODE
LOW	scripts/standalone/release.py	297	CODE
LOW	scripts/standalone/release.py	401	CODE
LOW	scripts/standalone/evidence.py	23	CODE
LOW	scripts/standalone/evidence.py	29	CODE
LOW	src/forge/server.py	1004	CODE
LOW	src/forge/clients/llamafile.py	215	CODE
LOW	src/forge/clients/llamafile.py	258	CODE
LOW	src/forge/clients/anthropic.py	261	CODE
LOW	src/forge/clients/anthropic.py	388	CODE
LOW	src/forge/clients/anthropic.py	558	CODE
LOW	src/forge/clients/ollama.py	37	CODE
LOW	src/forge/clients/ollama.py	399	CODE
LOW	src/forge/clients/base.py	218	CODE
LOW	src/forge/context/strategies.py	248	CODE
LOW	src/forge/core/steps.py	53	CODE
LOW	src/forge/core/runner.py	89	CODE
LOW	src/forge/core/workflow.py	68	CODE
LOW	src/forge/proxy/server.py	320	CODE
LOW	src/forge/proxy/server.py	530	CODE
LOW	src/forge/proxy/server.py	792	CODE
LOW	src/forge/proxy/convert.py	17	CODE
LOW	src/forge/proxy/convert_anthropic.py	20	CODE
LOW	src/forge/proxy/_profiles.py	60	CODE
LOW	src/forge/proxy/_installer.py	815	CODE
LOW	src/forge/proxy/__main__.py	152	CODE
LOW	src/forge/prompts/templates.py	12	CODE
LOW	src/forge/prompts/templates.py	55	CODE
LOW	src/forge/prompts/templates.py	185	CODE
LOW	src/forge/prompts/templates.py	238	CODE

Excessive Try-Catch Wrapping34 hits · 35 pts

Severity	File	Line	Snippet	Context
LOW⚡	tests/eval/batch_eval.py	961	except Exception:	CODE
LOW⚡	tests/eval/batch_eval.py	971	except Exception as exc:	CODE
LOW	tests/eval/dataset_builder.py	709	except Exception as exc:	STRING
LOW	tests/eval/dataset_builder.py	950	except Exception as exc:	STRING
LOW	tests/eval/dataset_builder.py	989	except Exception as exc:	CODE
MEDIUM	tests/eval/dataset_builder.py	91	def _require_pyarrow() -> tuple[Any, Any]:	CODE
LOW	tests/eval/eval_runner.py	218	except Exception as exc:	CODE
LOW	tests/eval/eval_runner.py	223	except Exception as exc:	CODE
LOW	tests/eval/eval_runner.py	377	except Exception as exc:	CODE
LOW	tests/eval/eval_runner.py	674	except Exception as exc:	CODE
LOW	docs/ARCHITECTURE.md	27	except Exception:	CODE
LOW	scripts/integration_test_proxy.py	291	except Exception as exc:	CODE
LOW	scripts/integration_test_proxy.py	839	except Exception as exc:	CODE
LOW	scripts/run_ablation.py	170	except Exception as e:	CODE
LOW	scripts/smoke_test_proxy.py	161	except Exception as exc: # make fixture bugs visible at the HTTP boundary	CODE
LOW	scripts/smoke_test_proxy.py	318	except Exception:	CODE
LOW	scripts/standalone/lifecycle_smoke.py	263	except Exception:	CODE
LOW	src/forge/server.py	451	except Exception as exc:	CODE
LOW	src/forge/server.py	459	except Exception as exc:	CODE
LOW	src/forge/server.py	559	except Exception:	CODE
LOW	src/forge/core/runner.py	346	except Exception as exc:	CODE
LOW	src/forge/core/slot_worker.py	137	except Exception as exc:	CODE
LOW	src/forge/proxy/server.py	269	except Exception:	CODE
LOW	src/forge/proxy/server.py	314	except Exception as exc:	CODE
LOW⚡	src/forge/proxy/server.py	390	except Exception:	CODE
LOW⚡	src/forge/proxy/server.py	394	except Exception:	CODE
LOW⚡	src/forge/proxy/server.py	400	except Exception:	CODE
LOW	src/forge/proxy/server.py	640	except Exception:	CODE
LOW	src/forge/proxy/server.py	789	except Exception as exc:	CODE
LOW	src/forge/proxy/server.py	882	except Exception as exc:	CODE
LOW	src/forge/proxy/_installer.py	896	except Exception:	CODE
LOW	src/forge/proxy/__main__.py	347	except Exception as exc:	STRING
LOW	src/forge/proxy/__main__.py	370	except Exception as exc:	STRING
LOW	src/forge/proxy/__main__.py	389	except Exception as exc:	STRING

Cross-Language Confusion6 hits · 30 pts

Severity	File	Line	Snippet	Context
HIGH	tests/eval/dataset_builder.py	696	if not parquet_file.schema_arrow.equals(schema, check_metadata=True):	STRING
HIGH⚡	src/forge/proxy/_installer.py	760	'while kill -0 "$parent" 2>/dev/null; do sleep 0.1; done',	CODE
HIGH⚡	src/forge/proxy/_installer.py	763	'[ "$(cat "$marker" 2>/dev/null)" = "$expected" ] \|\| exit 2',	CODE
HIGH	src/forge/proxy/_installer.py	788	f"rmdir -- {q(str(paths.command_dir))} 2>/dev/null \|\| true",	CODE
HIGH	src/forge/proxy/_installer.py	789	f"rmdir -- {q(str(paths.root))} 2>/dev/null \|\| true",	CODE
HIGH	src/forge/prompts/nudges.py	69	args content (e.g. an empty string, null, a list, or a primitive	STRING

Structural Annotation Overuse16 hits · 28 pts

Severity	File	Line	Snippet	Context
LOW⚡	tests/eval/scenarios/_compaction_chain.py	532	# Step 1: patient MRN (not in user message)	COMMENT
LOW⚡	tests/eval/scenarios/_compaction_chain.py	534	# Step 3: lab values (only from tool result)	COMMENT
LOW⚡	tests/eval/scenarios/_compaction_chain.py	536	# Step 4: imaging findings (only from tool result)	COMMENT
LOW⚡	tests/eval/scenarios/_compaction_chain.py	538	# Step 5: specialist (only from tool result)	COMMENT
LOW⚡	tests/eval/scenarios/_compaction_chain.py	540	# Step 6: medication concern (only from tool result)	COMMENT
LOW	docs/decisions/003-thinking-label-ux.md	64	### Step 1: Add `_resolve_reasoning()` to OllamaClient	COMMENT
LOW	docs/decisions/003-thinking-label-ux.md	86	### Step 2: Use it in `send()` (batch path)	COMMENT
LOW	docs/decisions/003-thinking-label-ux.md	102	### Step 3: Use it in `_iter_stream()` (streaming path)	COMMENT
LOW	docs/decisions/003-thinking-label-ux.md	124	### Step 4: Add unit tests	COMMENT
LOW	docs/decisions/003-thinking-label-ux.md	187	### Step 5: Update README roadmap	COMMENT
LOW	docs/decisions/004-async-on-chunk.md	75	### Step 1: Change type signature	COMMENT
LOW	docs/decisions/004-async-on-chunk.md	91	### Step 2: Await the callback in `_send_streaming()`	COMMENT
LOW	docs/decisions/004-async-on-chunk.md	107	### Step 3: Update `_make_runner` helper type hint	COMMENT
LOW	docs/decisions/004-async-on-chunk.md	127	### Step 4: Update the test	COMMENT
LOW	docs/decisions/004-async-on-chunk.md	171	### Step 5: Update docstring	COMMENT
LOW	docs/decisions/004-async-on-chunk.md	180	### Step 6: Update README roadmap	COMMENT

Docstring Block Structure5 hits · 25 pts

Severity	File	Line	Snippet	Context
HIGH	src/forge/server.py	656	Resolve the ContextManager budget for the given mode. Args: mode: The budget mode to use.	STRING
HIGH	src/forge/server.py	711	Start server with the specified budget mode and return the resolved budget. Handles the mode-specific startup d	STRING
HIGH	src/forge/clients/sampling_defaults.py	254	Apply the recommended-sampling policy for ``model``. Called by client constructors at instantiation time. The four-	STRING
HIGH	src/forge/core/inference.py	200	Send messages to the LLM with compaction, folding, validation, and retry. Retries are handled internally — the call	STRING
HIGH	src/forge/core/slot_worker.py	65	Submit a workflow for execution and wait for the result. Args: workflow: The workflow to execute.	STRING

Fake / Example Data18 hits · 21 pts

Severity	File	Line	Snippet	Context
LOW	tests/unit/test_argument_transformation.py	68	"Acme Corp",	CODE
LOW	tests/unit/test_argument_transformation.py	87	{"vendor_name": "Acme Corp"},	CODE
LOW	tests/unit/test_argument_transformation.py	237	assert "Acme Corp" in system.get_approved_vendors()	CODE
LOW⚡	tests/unit/test_inconsistent_api_recovery.py	38	assert "Acme Corp" in out	CODE
LOW⚡	tests/unit/test_inconsistent_api_recovery.py	261	assert "Acme Corp" in out	CODE
LOW	tests/eval/scenarios/_base.py	15	placeholder_field: str = Field(description="placeholder")	CODE
LOW⚡	tests/eval/scenarios/_base.py	49	spec=ToolSpec(name=terminal_tool, description="placeholder",	CODE
LOW⚡	tests/eval/scenarios/_base.py	56	spec=ToolSpec(name=s, description="placeholder",	CODE
LOW⚡	tests/eval/scenarios/_base.py	64	description="placeholder",	CODE
LOW	tests/eval/scenarios/_model_reasoning.py	300	# approved 'Acme Corp'. Literal string compare over-flags it;	COMMENT
LOW	tests/eval/scenarios/_model_reasoning.py	313	"Acme Corp",	CODE
LOW	tests/eval/scenarios/_model_reasoning.py	326	("TX-1002", "2024-10-12", "Acme Corp", 12000, "USD", False), # approved	CODE
LOW	tests/eval/scenarios/_model_reasoning.py	377	if name == "Acme Corp":	CODE
LOW	tests/eval/scenarios/_model_reasoning.py	621	(12345, "Acme Corp"),	CODE
LOW⚡	tests/eval/scenarios/_stateful_model_reasoning.py	368	"Acme Corp",	CODE
LOW⚡	tests/eval/scenarios/_stateful_model_reasoning.py	378	("TX-1002", "2024-10-12", "Acme Corp", 12000, "USD"),	CODE
LOW	tests/eval/scenarios/_stateful_model_reasoning.py	441	if name == "Acme Corp":	CODE
LOW	tests/eval/scenarios/_stateful_model_reasoning.py	681	(12345, "Acme Corp"),	CODE

Cross-File Repetition4 hits · 20 pts

Severity	File	Snippet	Context
HIGH	src/forge/clients/vllm.py	per-call headers to apply, enforcing the one-credential rule. returns the dict to pass as httpx ``headers=`` (merged ove	STRING
HIGH	src/forge/clients/llamafile.py	per-call headers to apply, enforcing the one-credential rule. returns the dict to pass as httpx ``headers=`` (merged ove	STRING
HIGH	src/forge/clients/openai_compat.py	per-call headers to apply, enforcing the one-credential rule. returns the dict to pass as httpx ``headers=`` (merged ove	STRING
HIGH	src/forge/clients/ollama.py	per-call headers to apply, enforcing the one-credential rule. returns the dict to pass as httpx ``headers=`` (merged ove	STRING

Over-Commented Block19 hits · 18 pts

Severity	File	Line	Snippet	Context
LOW	install.sh	1	#!/bin/sh	COMMENT
LOW	tests/eval/report.py	501	# model under different identity strings:	COMMENT
LOW	tests/eval/scenarios/_model_reasoning.py	61	top_vendor: str = Field(description="Vendor of the single largest flagged transaction")	COMMENT
LOW	tests/eval/scenarios/_model_reasoning.py	281	tags=["advanced_reasoning", "reasoning", "model_quality"],	COMMENT
LOW	tests/eval/scenarios/_model_reasoning.py	561		COMMENT
LOW	tests/eval/scenarios/_model_reasoning.py	881	"txn_count (int)."	COMMENT
LOW	tests/eval/scenarios/_model_reasoning.py	901	# Sarah Chen — top-decile attractor; BLOCKER: Stripe non-compete	COMMENT
LOW	examples/foreign_loop.py	141	# The respond tool eliminates this ambiguity: the model calls	COMMENT
LOW	src/forge/clients/sampling_defaults.py	101	"gemma4:e4b-it-q4_K_M": {"temperature": 1.0, "top_p": 0.95, "top_k": 64}, # https://huggingface.co/go	COMMENT
LOW	src/forge/clients/sampling_defaults.py	121	# supports low/high/max reasoning effort. This one registry value is the	COMMENT
LOW	src/forge/clients/sampling_defaults.py	141	# so the model emits something substantive instead of empty <think> blocks. Native tool format is	COMMENT
LOW	src/forge/clients/sampling_defaults.py	181	"ministral-3:8b-reasoning-2512-q8_0": {"temperature": 0.7}, # https://huggingface.co/mistralai/Ministral-3-8B-Re	COMMENT
LOW	src/forge/clients/sampling_defaults.py	201	"Mellum2-12B-A2.5B-Instruct-Q4_K_M": {"temperature": 0.6, "top_p": 0.95, "top_k": 20}, # https://huggingface.co/	COMMENT
LOW	src/forge/clients/llamafile.py	41	# index is filesystem layout, not model identity, so strip it for the	COMMENT
LOW	src/forge/clients/llamafile.py	61	# The gate is STRUCTURAL, not phrase-based: a 500 is rescuable iff its body	COMMENT
LOW	src/forge/clients/anthropic.py	41	# lowercases all inbound header keys, so a relocated ``x-api-key`` would fail	COMMENT
LOW	src/forge/clients/anthropic.py	121	) -> None:	COMMENT
LOW	src/forge/clients/anthropic.py	141	log.debug(	COMMENT
LOW	src/forge/guardrails/nudge.py	41		COMMENT

AI Structural Patterns18 hits · 16 pts

Severity	File	Line	Context
LOW	tests/unit/test_runner.py	69	CODE
LOW	tests/unit/test_proxy_server.py	143	CODE
LOW	tests/eval/batch_eval.py	1065	CODE
LOW	src/forge/server.py	1004	CODE
LOW	src/forge/server.py	1106	CODE
LOW	src/forge/server.py	236	CODE
LOW	src/forge/server.py	695	CODE
LOW	src/forge/clients/vllm.py	67	CODE
LOW	src/forge/clients/llamafile.py	332	CODE
LOW	src/forge/clients/openai_compat.py	61	CODE
LOW	src/forge/clients/anthropic.py	108	CODE
LOW	src/forge/clients/ollama.py	106	CODE
LOW	src/forge/core/runner.py	35	CODE
LOW	src/forge/core/inference.py	179	CODE
LOW	src/forge/proxy/server.py	180	CODE
LOW	src/forge/proxy/handler.py	256	CODE
LOW	src/forge/proxy/proxy.py	62	CODE
LOW	src/forge/proxy/_installer.py	924	CODE

Modern Structural Boilerplate15 hits · 15 pts

Severity	File	Line	Snippet	Context
LOW	tests/eval/scenarios/__init__.py	81	__all__ = [	CODE
LOW	scripts/integration_test_proxy.py	98	def _setup_logging() -> None:	CODE
LOW	src/forge/server.py	221	def _set_resolved_daemon_target(self, target: ResolvedBackend) -> None:	CODE
LOW	src/forge/__init__.py	78	__all__ = [	CODE
LOW	src/forge/clients/vllm.py	173	def _set_model_identity(self, wire_id: str \| Path) -> None:	CODE
LOW	src/forge/clients/__init__.py	14	__all__ = [	CODE
LOW	src/forge/clients/ollama.py	462	def set_num_ctx(self, num_ctx: int \| None) -> None:	CODE
LOW	src/forge/tools/__init__.py	9	__all__ = [	CODE
LOW	src/forge/context/__init__.py	20	__all__ = [	CODE
LOW	src/forge/context/hardware.py	21	logger = logging.getLogger(__name__)	CODE
LOW	src/forge/context/manager.py	131	def update_token_count(self, total_tokens: int) -> None:	CODE
LOW	src/forge/core/__init__.py	3	__all__ = ["WorkflowRunner"]	CODE
LOW	src/forge/guardrails/__init__.py	19	__all__ = [	CODE
LOW	src/forge/proxy/__init__.py	5	__all__ = ["ProxyServer"]	CODE
LOW	src/forge/prompts/__init__.py	6	__all__ = [	CODE

Verbosity Indicators5 hits · 11 pts

Severity	File	Line	Snippet	Context
LOW⚡	tests/eval/scenarios/_compaction_chain.py	532	# Step 1: patient MRN (not in user message)	COMMENT
LOW⚡	tests/eval/scenarios/_compaction_chain.py	534	# Step 3: lab values (only from tool result)	COMMENT
LOW⚡	tests/eval/scenarios/_compaction_chain.py	536	# Step 4: imaging findings (only from tool result)	COMMENT
LOW⚡	tests/eval/scenarios/_compaction_chain.py	538	# Step 5: specialist (only from tool result)	COMMENT
LOW⚡	tests/eval/scenarios/_compaction_chain.py	540	# Step 6: medication concern (only from tool result)	COMMENT

AI Slop Vocabulary1 hit · 4 pts

Severity	File	Line	Snippet	Context
MEDIUM⚡	tests/unit/test_proxy_auth.py	45	# The proxy lowercases keys, but be robust to mixed case anyway.	COMMENT

Redundant / Tautological Comments1 hit · 2 pts

Severity	File	Line	Snippet	Context
LOW	examples/foreign_loop.py	175	# Check if the model called respond -- extract the message	COMMENT

Analysis Overview

What These Metrics Mean

Score History

Severity Breakdown

Directory Score Breakdown

Pattern Findings