Skip to content

armgit / letta

Public Imported from GitHub
0

Platform for stateful agents: AI with advanced memory that can learn and self-improve over time.

fix: Change to only derive json_schema for custom tools if missing (#3129)

Matthew Zhou802361493fee
diff --git a/letta/schemas/tool.py b/letta/schemas/tool.py
index 47d9d6e8b..844ee77ef 100644
--- a/letta/schemas/tool.py
+++ b/letta/schemas/tool.py
@@ -91,7 +91,8 @@ class Tool(BaseTool):
description=description,
append_heartbeat=False,
)
- else:
+ elif not self.json_schema:
+ # If there's not a json_schema provided, then we need to re-derive
try:
self.json_schema = derive_openai_json_schema(source_code=self.source_code)
except Exception as e:
diff --git a/letta/services/file_processor/embedder/openai_embedder.py b/letta/services/file_processor/embedder/openai_embedder.py
index 7e6fa55e4..b831c819c 100644
--- a/letta/services/file_processor/embedder/openai_embedder.py
+++ b/letta/services/file_processor/embedder/openai_embedder.py
@@ -44,7 +44,7 @@ class OpenAIEmbedder:
{
"batch_size": len(batch),
"model": self.embedding_config.embedding_model,
- "provider": self.embedding_config.embedding_provider,
+ "embedding_endpoint_type": self.embedding_config.embedding_endpoint_type,
},
)
embeddings = await self.client.request_embeddings(inputs=batch, embedding_config=self.embedding_config)
@@ -63,7 +63,7 @@ class OpenAIEmbedder:
{
"total_chunks": len(chunks),
"model": self.embedding_config.embedding_model,
- "provider": self.embedding_config.embedding_provider,
+ "embedding_endpoint_type": self.embedding_config.embedding_endpoint_type,
"batch_size": self.embedding_config.batch_size,
"file_id": file_id,
"source_id": source_id,
diff --git a/tests/test_agent_serialization.py b/tests/test_agent_serialization.py
index 000db3b98..a5ad070c2 100644
--- a/tests/test_agent_serialization.py
+++ b/tests/test_agent_serialization.py
@@ -239,7 +239,7 @@ def strip_datetime_fields(d: Dict[str, Any]) -> Dict[str, Any]:
return {k: v for k, v in d.items() if not isinstance(v, datetime)}
-def _log_mismatch(key: str, expected: Any, actual: Any, log: bool) -> None:
+def _log_mismatch(key: str, expected: Any, actual: Any, log: bool = True) -> None:
"""Log detailed information about a mismatch."""
if log:
print(f"\n🔴 Mismatch Found in Key: '{key}'")
@@ -257,6 +257,8 @@ def _compare_agent_state_model_dump(d1: Dict[str, Any], d2: Dict[str, Any], log:
Compare two dictionaries with special handling:
- Keys in `ignore_prefix_fields` should match only by prefix.
- 'message_ids' lists should match in length only.
+ - 'tool_exec_environment_variables' ignores values since they're cleared during serialization.
+ - 'json_schema' allows extra fields in d2 (from schema normalization during deserialization).
- Datetime fields are ignored.
- Order-independent comparison for lists of dicts.
"""
@@ -266,6 +268,21 @@ def _compare_agent_state_model_dump(d1: Dict[str, Any], d2: Dict[str, Any], log:
d1 = strip_datetime_fields(d1)
d2 = strip_datetime_fields(d2)
+ # For json_schema comparison, allow d2 to have extra fields (schema normalization)
+ if "json_schema" in d1 and "json_schema" in d2:
+ schema1 = d1["json_schema"]
+ schema2 = d2["json_schema"]
+ if isinstance(schema1, dict) and isinstance(schema2, dict):
+ # Check that all fields in schema1 exist in schema2 with same values
+ for k, v in schema1.items():
+ if k not in schema2 or not _compare_agent_state_model_dump({k: v}, {k: schema2[k]}, log=False):
+ _log_mismatch("json_schema", schema1, schema2, log)
+ return False
+ # Don't compare other keys for json_schema
+ d1_without_schema = {k: v for k, v in d1.items() if k != "json_schema"}
+ d2_without_schema = {k: v for k, v in d2.items() if k != "json_schema"}
+ return _compare_agent_state_model_dump(d1_without_schema, d2_without_schema, log)
+
if d1.keys() != d2.keys():
_log_mismatch("dict_keys", set(d1.keys()), set(d2.keys()))
return False
@@ -285,8 +302,17 @@ def _compare_agent_state_model_dump(d1: Dict[str, Any], d2: Dict[str, Any], log:
if not isinstance(v1, list) or not isinstance(v2, list) or len(v1) != len(v2):
_log_mismatch(key, v1, v2, log)
return False
+ # Compare environment variables ignoring values (cleared during serialization)
+ for env1, env2 in zip(v1, v2):
+ if isinstance(env1, dict) and isinstance(env2, dict):
+ # Compare all fields except 'value'
+ env1_without_value = {k: v for k, v in env1.items() if k != "value"}
+ env2_without_value = {k: v for k, v in env2.items() if k != "value"}
+ if not _compare_agent_state_model_dump(env1_without_value, env2_without_value, log=False):
+ _log_mismatch(key, v1, v2, log)
+ return False
elif isinstance(v1, Dict) and isinstance(v2, Dict):
- if not _compare_agent_state_model_dump(v1, v2):
+ if not _compare_agent_state_model_dump(v1, v2, log=False):
_log_mismatch(key, v1, v2, log)
return False
elif isinstance(v1, list) and isinstance(v2, list):
diff --git a/tests/test_sources.py b/tests/test_sources.py
index c1aa9972e..ed8814174 100644
--- a/tests/test_sources.py
+++ b/tests/test_sources.py
@@ -681,8 +681,8 @@ def test_open_files_schema_descriptions(client: LettaSDKClient):
# Check that examples are included
assert "Examples:" in description
- assert 'FileOpenRequest(file_name="config.py")' in description
- assert 'FileOpenRequest(file_name="config.py", offset=1, length=50)' in description
+ assert 'FileOpenRequest(file_name="project_utils/config.py")' in description
+ assert 'FileOpenRequest(file_name="project_utils/config.py", offset=1, length=50)' in description
assert "# Lines 1-50" in description
assert "# Lines 100-199" in description
assert "# Entire file" in description