fix: tracing count, TS evaluator template, revision data validation, …

api/oss/src/apis/fastapi/tracing/router.py

@@ -301,7 +301,7 @@ async def ingest_spans(  # MUTATION
         )
 
         link_response = OTelLinksResponse(
-            count=len(links) + len(dropped),
+            count=len(links),
             links=links,
             dropped=dropped or None,
         )

api/oss/src/core/queries/dtos.py

@@ -1,7 +1,7 @@
 from typing import List, Optional
 from uuid import UUID
 
-from pydantic import BaseModel, Field
+from pydantic import BaseModel, ConfigDict, Field
 
 
 from oss.src.core.tracing.dtos import Filtering, Formatting
@@ -119,6 +119,8 @@ class QueryVariantQuery(VariantQuery):
 
 
 class QueryRevisionData(BaseModel):
+    model_config = ConfigDict(extra="forbid")
+
     formatting: Optional[Formatting] = None
     filtering: Optional[Filtering] = None
     windowing: Optional[Windowing] = None

api/oss/src/core/shared/exceptions.py

@@ -18,7 +18,8 @@ def __init__(
     def __str__(self):
         _message = self.message
 
-        for key, value in self.conflict.items():
-            _message += f" {key}={value}"
+        if self.conflict:
+            for key, value in self.conflict.items():
+                _message += f" {key}={value}"
 
         return _message

api/oss/src/core/testsets/dtos.py

@@ -1,7 +1,7 @@
 from typing import Optional, List, Tuple
 from uuid import UUID
 
-from pydantic import BaseModel, Field
+from pydantic import BaseModel, ConfigDict, Field
 
 from oss.src.core.shared.dtos import (
     sync_alias,
@@ -134,6 +134,8 @@ class TestsetVariantQuery(VariantQuery):
 
 
 class TestsetRevisionData(BaseModel):
+    model_config = ConfigDict(extra="forbid")
+
     testcase_ids: Optional[List[UUID]] = None
     testcases: Optional[List[Testcase]] = None
 

docs/docs/reference/api-guide/10-tracing.mdx

@@ -56,7 +56,7 @@ Each span also declares its role through `ag.type.span`:
 | `chat`, `completion` | Chat-completion or completion calls. |
 | `embedding`, `query`, `rerank` | Index and retrieval calls. |
 
-## References
+## References {#references-and-entity-linking}
 
 A reference links a trace (or a span) back to the Agenta entity that produced or triggered it. References live under `ag.references.<key>` and each value names an entity revision:
 
@@ -125,7 +125,7 @@ The Agenta-native ingest endpoints `POST /tracing/spans/ingest` and `POST /trace
 
 :::
 
-### How ingest works
+### How ingest works {#async-write-contract-202}
 
 Every ingest endpoint returns `202 Accepted` *before* the spans are persisted. The router parses the payload, hands the spans off to an async worker, and responds immediately. This keeps the ingest buffer fast, so instrumented apps do not block on telemetry.
 

sdks/python/agenta/sdk/engines/running/templates.py

@@ -66,13 +66,11 @@
 {user_code}
 
 // Execute and capture result
-let result = evaluate(app_params, inputs, output, correct_answer);
+let result: number | null = evaluate(app_params, inputs, output, correct_answer);
 
 // Ensure result is a number
-result = Number(result);
-if (!Number.isFinite(result)) {{
-    result = null;
-}}
+const resultNum = Number(result);
+result = Number.isFinite(resultNum) ? resultNum : null;
 
 // Print result for capture
 console.log(JSON.stringify({{ result: result }}));
@@ -137,13 +135,11 @@
 {user_code}
 
 // Execute and capture result
-let result = evaluate(inputs, outputs, trace);
+let result: number | null = evaluate(inputs, outputs, trace);
 
 // Ensure result is a number
-result = Number(result);
-if (!Number.isFinite(result)) {{
-    result = null;
-}}
+const resultNum = Number(result);
+result = Number.isFinite(resultNum) ? resultNum : null;
 
 // Print result for capture
 console.log(JSON.stringify({{ result: result }}));

sdks/python/agenta/sdk/models/workflows.py

@@ -119,6 +119,8 @@ class WorkflowQueryFlags(BaseModel):
 
 
 class WorkflowRevisionData(BaseModel):
+    model_config = ConfigDict(extra="forbid")
+
     uri: Optional[str] = None
 
     url: Optional[str] = None