refactor: clean up fresh tech debt from 2026-09-30 (#43993)

* refactor: clean up fresh tech debt from 2026-09-30

Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>

* refactor: keep agent tracing route comment in LiteLLMRoutes

Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>

* refactor: move agent tracing route comment above the trace routes

Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>

* refactor: drop route comment that duplicates the trace handler docstring

Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>

---------

Co-authored-by: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>

litellm/constants.py

@@ -46,7 +46,6 @@ ROUTER_SETTINGS_MANAGED_OUTSIDE_CONFIG: Final[frozenset[str]] = frozenset(
 )
 DEFAULT_BATCH_SIZE: Final = int(os.getenv("DEFAULT_BATCH_SIZE", 512))
 DEFAULT_FLUSH_INTERVAL_SECONDS: Final = int(os.getenv("DEFAULT_FLUSH_INTERVAL_SECONDS", 5))
-# Agent tracing / ClickHouse
 CLICKHOUSE_BATCH_SIZE: Final = get_env_int("CLICKHOUSE_BATCH_SIZE", 10_000)
 CLICKHOUSE_FLUSH_INTERVAL_SECONDS: Final = float(os.getenv("CLICKHOUSE_FLUSH_INTERVAL_SECONDS", "1.0"))
 CLICKHOUSE_MAX_BUFFERED_ROWS: Final = get_env_int("CLICKHOUSE_MAX_BUFFERED_ROWS", 200_000)

litellm/proxy/_types.py

@@ -520,7 +520,6 @@ class LiteLLMRoutes(enum.Enum):
         "/v1/rag/ingest",
         "/rag/query",
         "/v1/rag/query",
-        # agent tracing: OTLP ingest + reads (scoped to the caller's team in the handler)
         "/engine",
         "/engine/{engine_id}",
         "/engine/{engine_id}/runs",

litellm/proxy/roi_calculator/estimator.py

@@ -92,32 +92,6 @@ def cache_context(settings: ROISettings, models: tuple[EstimatorModel, ...] | No
     return hashlib.sha256(context.encode()).hexdigest()
 
 
-def pull_cache_key(
-    settings: ROISettings,
-    pull: ROIPullEvidence,
-    models: tuple[EstimatorModel, ...] | None = None,
-) -> str:
-    evidence: Final = json.dumps(
-        metadata_evidence(pull).model_dump(exclude_unset=True),
-        ensure_ascii=False,
-    )
-    key: Final = json.dumps(
-        (
-            ESTIMATE_VERSION,
-            settings.estimator_model,
-            settings.estimator_prompt,
-            RESPONSE_CONTRACT,
-            estimator_options(_configured_models(settings, models)),
-            pull["repo"],
-            pull["number"],
-            pull["head_sha"],
-            evidence,
-        ),
-        ensure_ascii=False,
-    )
-    return hashlib.sha256(key.encode()).hexdigest()
-
-
 class Estimator:
     def __init__(
         self,

litellm/proxy/roi_calculator/pull_cache.py

@@ -2,7 +2,6 @@ import hashlib
 import json
 from typing import Final
 
-from litellm.proxy.roi_calculator.estimator import cache_context
 from litellm.proxy.roi_calculator.github import GitHubPullListItem
 from litellm.types.roi_calculator import ROISettings
 
@@ -46,7 +45,3 @@ def settings_fingerprint(settings: ROISettings) -> str:
         ensure_ascii=False,
     )
     return hashlib.sha256(value.encode()).hexdigest()
-
-
-def current_cache_context(settings: ROISettings) -> str:
-    return cache_context(settings)

litellm/tracing/decode.py

@@ -52,9 +52,6 @@ class OTLPPayloadTooLargeError(OverflowError):
     pass
 
 
-# ---------------------------------------------------------------- decode
-
-
 def _truncate(value: str) -> str:
     size = len(value.encode("utf-8"))
     if size <= OTLP_MAX_ATTRIBUTE_VALUE_BYTES:
@@ -123,9 +120,6 @@ def _span_row(span: DecodedSpan) -> SpanRow:
     return row
 
 
-# ---------------------------------------------------------------- normalize
-
-
 def _loads(value: str) -> object:
     try:
         return json.loads(value)

litellm/tracing/receiver.py

@@ -82,8 +82,6 @@ class TraceReceiver:
             spend_log_retention_days=AGENT_TRACING_SPEND_LOG_RETENTION_DAYS,
         )
 
-    # ------------------------------------------------------------ write
-
     async def ingest(
         self,
         body: bytes,
@@ -108,8 +106,6 @@ class TraceReceiver:
             raise TracingPayloadTooLargeError(str(error)) from error
         return len(rows)
 
-    # ------------------------------------------------------------ read
-
     async def list_traces(self, scope: TraceScope, start_ms: int, end_ms: int, cursor: str | None = None) -> TracePage:
         return await self.store.list_traces(scope, start_ms, end_ms, cursor)