feat: Add custom agent parameters (llm, iteration_budget, include_research) (#96)

Add support for the new agent endpoint parameters from the server:

- `llm`: Specific LLM model (LLMEnumPublic enum with CLAUDE_4_5_HAIKU, GPT_5, etc.)
- `iteration_budget`: Number of agent iterations (0-20)
- `include_research`: Include research notes in response

These can be used instead of `effort_level` presets by setting `effort_level=None`
and providing all three custom parameters.

Changes:
- Regenerate OpenAPI client with new LLMEnumPublic enum
- Update single_agent and agent_map to accept new parameters
- Add tests verifying correct parameter passing to API

Co-authored-by: Claude Opus 4.5 <noreply@anthropic.com>

Author: RafaelPo

src/everyrow/generated/api/artifacts/__init__.py

@@ -0,0 +1 @@
+"""Contains endpoint functions for accessing the API"""

src/everyrow/generated/api/billing/__init__.py

@@ -0,0 +1 @@
+"""Contains endpoint functions for accessing the API"""

src/everyrow/generated/api/operations/__init__.py

@@ -0,0 +1 @@
+"""Contains endpoint functions for accessing the API"""

src/everyrow/generated/api/operations/agent_map_operations_agent_map_post.py

@@ -75,6 +75,12 @@ def sync_detailed(
 
      Run an AI agent on each row in parallel to perform research and generate responses.
 
+    **Configuration options** (mutually exclusive):
+
+    1. **Use a preset** - set `effort_level` to `low`, `medium`, or `high`
+    2. **Fully customize** - set `effort_level=null` and provide ALL of: `llm`, `iteration_budget`,
+    `include_research`
+
     Args:
         body (AgentMapOperation):
 
@@ -106,6 +112,12 @@ def sync(
 
      Run an AI agent on each row in parallel to perform research and generate responses.
 
+    **Configuration options** (mutually exclusive):
+
+    1. **Use a preset** - set `effort_level` to `low`, `medium`, or `high`
+    2. **Fully customize** - set `effort_level=null` and provide ALL of: `llm`, `iteration_budget`,
+    `include_research`
+
     Args:
         body (AgentMapOperation):
 
@@ -132,6 +144,12 @@ async def asyncio_detailed(
 
      Run an AI agent on each row in parallel to perform research and generate responses.
 
+    **Configuration options** (mutually exclusive):
+
+    1. **Use a preset** - set `effort_level` to `low`, `medium`, or `high`
+    2. **Fully customize** - set `effort_level=null` and provide ALL of: `llm`, `iteration_budget`,
+    `include_research`
+
     Args:
         body (AgentMapOperation):
 
@@ -161,6 +179,12 @@ async def asyncio(
 
      Run an AI agent on each row in parallel to perform research and generate responses.
 
+    **Configuration options** (mutually exclusive):
+
+    1. **Use a preset** - set `effort_level` to `low`, `medium`, or `high`
+    2. **Fully customize** - set `effort_level=null` and provide ALL of: `llm`, `iteration_budget`,
+    `include_research`
+
     Args:
         body (AgentMapOperation):
 

src/everyrow/generated/api/operations/single_agent_operations_single_agent_post.py

@@ -75,6 +75,20 @@ def sync_detailed(
 
      Run a single AI agent to perform research and generate a response.
 
+    **Configuration options** (mutually exclusive):
+
+    1. **Use a preset** - set `effort_level` to one of:
+       - `low`: Fast, minimal research (Gemini Flash, 0 iterations, no provenance)
+       - `medium`: Balanced (Gemini Flash High, 5 iterations, with provenance)
+       - `high`: Thorough research (Claude Opus, 10 iterations, with provenance)
+
+    2. **Fully customize** - set `effort_level=null` and provide ALL of:
+       - `llm`: The LLM model to use
+       - `iteration_budget`: Number of agent iterations (0-20)
+       - `include_research`: Whether to include research notes
+
+    You cannot mix these approaches - either use a preset OR specify all custom parameters.
+
     Args:
         body (SingleAgentOperation):
 
@@ -106,6 +120,20 @@ def sync(
 
      Run a single AI agent to perform research and generate a response.
 
+    **Configuration options** (mutually exclusive):
+
+    1. **Use a preset** - set `effort_level` to one of:
+       - `low`: Fast, minimal research (Gemini Flash, 0 iterations, no provenance)
+       - `medium`: Balanced (Gemini Flash High, 5 iterations, with provenance)
+       - `high`: Thorough research (Claude Opus, 10 iterations, with provenance)
+
+    2. **Fully customize** - set `effort_level=null` and provide ALL of:
+       - `llm`: The LLM model to use
+       - `iteration_budget`: Number of agent iterations (0-20)
+       - `include_research`: Whether to include research notes
+
+    You cannot mix these approaches - either use a preset OR specify all custom parameters.
+
     Args:
         body (SingleAgentOperation):
 
@@ -132,6 +160,20 @@ async def asyncio_detailed(
 
      Run a single AI agent to perform research and generate a response.
 
+    **Configuration options** (mutually exclusive):
+
+    1. **Use a preset** - set `effort_level` to one of:
+       - `low`: Fast, minimal research (Gemini Flash, 0 iterations, no provenance)
+       - `medium`: Balanced (Gemini Flash High, 5 iterations, with provenance)
+       - `high`: Thorough research (Claude Opus, 10 iterations, with provenance)
+
+    2. **Fully customize** - set `effort_level=null` and provide ALL of:
+       - `llm`: The LLM model to use
+       - `iteration_budget`: Number of agent iterations (0-20)
+       - `include_research`: Whether to include research notes
+
+    You cannot mix these approaches - either use a preset OR specify all custom parameters.
+
     Args:
         body (SingleAgentOperation):
 
@@ -148,7 +190,6 @@ async def asyncio_detailed(
     )
 
     response = await client.get_async_httpx_client().request(**kwargs)
-
     return _build_response(client=client, response=response)
 
 
@@ -161,6 +202,20 @@ async def asyncio(
 
      Run a single AI agent to perform research and generate a response.
 
+    **Configuration options** (mutually exclusive):
+
+    1. **Use a preset** - set `effort_level` to one of:
+       - `low`: Fast, minimal research (Gemini Flash, 0 iterations, no provenance)
+       - `medium`: Balanced (Gemini Flash High, 5 iterations, with provenance)
+       - `high`: Thorough research (Claude Opus, 10 iterations, with provenance)
+
+    2. **Fully customize** - set `effort_level=null` and provide ALL of:
+       - `llm`: The LLM model to use
+       - `iteration_budget`: Number of agent iterations (0-20)
+       - `include_research`: Whether to include research notes
+
+    You cannot mix these approaches - either use a preset OR specify all custom parameters.
+
     Args:
         body (SingleAgentOperation):
 

src/everyrow/generated/api/sessions/__init__.py

@@ -0,0 +1 @@
+"""Contains endpoint functions for accessing the API"""

src/everyrow/generated/api/tasks/__init__.py

@@ -0,0 +1 @@
+"""Contains endpoint functions for accessing the API"""

src/everyrow/generated/models/__init__.py

@@ -17,14 +17,15 @@
 from .error_response_details_type_0 import ErrorResponseDetailsType0
 from .http_validation_error import HTTPValidationError
 from .insufficient_balance_error import InsufficientBalanceError
+from .llm_enum_public import LLMEnumPublic
 from .merge_operation import MergeOperation
 from .merge_operation_left_input_type_1_item import MergeOperationLeftInputType1Item
 from .merge_operation_left_input_type_2 import MergeOperationLeftInputType2
 from .merge_operation_right_input_type_1_item import MergeOperationRightInputType1Item
 from .merge_operation_right_input_type_2 import MergeOperationRightInputType2
+from .merge_operation_use_web_search_type_0 import MergeOperationUseWebSearchType0
 from .operation_response import OperationResponse
 from .public_effort_level import PublicEffortLevel
-from .public_llm import PublicLLM
 from .public_task_type import PublicTaskType
 from .rank_operation import RankOperation
 from .rank_operation_input_type_1_item import RankOperationInputType1Item
@@ -64,14 +65,15 @@
     "ErrorResponseDetailsType0",
     "HTTPValidationError",
     "InsufficientBalanceError",
+    "LLMEnumPublic",
     "MergeOperation",
     "MergeOperationLeftInputType1Item",
     "MergeOperationLeftInputType2",
     "MergeOperationRightInputType1Item",
     "MergeOperationRightInputType2",
+    "MergeOperationUseWebSearchType0",
     "OperationResponse",
     "PublicEffortLevel",
-    "PublicLLM",
     "PublicTaskType",
     "RankOperation",
     "RankOperationInputType1Item",

src/everyrow/generated/models/agent_map_operation.py

@@ -7,8 +7,8 @@
 from attrs import define as _attrs_define
 from attrs import field as _attrs_field
 
+from ..models.llm_enum_public import LLMEnumPublic
 from ..models.public_effort_level import PublicEffortLevel
-from ..models.public_llm import PublicLLM
 from ..types import UNSET, Unset
 
 if TYPE_CHECKING:
@@ -31,19 +31,28 @@ class AgentMapOperation:
         session_id (None | Unset | UUID): Session ID. If not provided, a new session is auto-created for this task.
         response_schema (AgentMapOperationResponseSchemaType0 | None | Unset): JSON Schema for the response format. If
             not provided, use default answer schema.
-        llm (PublicLLM | Unset):
-        effort_level (PublicEffortLevel | Unset):
+        llm (LLMEnumPublic | None | Unset): LLM to use for each agent. Required when effort_level is not set.
+        effort_level (None | PublicEffortLevel | Unset): Effort level preset: low (quick), medium (balanced), high
+            (thorough). Mutually exclusive with llm/iteration_budget/include_research - use either a preset or custom
+            params, not both. If not specified, you must provide all individual parameters (llm, iteration_budget,
+            include_research).
         join_with_input (bool | Unset): If True, merge agent output with input row. If False, output only agent results.
             Default: True.
+        iteration_budget (int | None | Unset): Number of agent iterations per row (0-20). Required when effort_level is
+            not set.
+        include_research (bool | None | Unset): Include research notes in the response. Required when effort_level is
+            not set.
     """
 
     input_: AgentMapOperationInputType2 | list[AgentMapOperationInputType1Item] | UUID
     task: str
     session_id: None | Unset | UUID = UNSET
     response_schema: AgentMapOperationResponseSchemaType0 | None | Unset = UNSET
-    llm: PublicLLM | Unset = UNSET
-    effort_level: PublicEffortLevel | Unset = UNSET
+    llm: LLMEnumPublic | None | Unset = UNSET
+    effort_level: None | PublicEffortLevel | Unset = UNSET
     join_with_input: bool | Unset = True
+    iteration_budget: int | None | Unset = UNSET
+    include_research: bool | None | Unset = UNSET
     additional_properties: dict[str, Any] = _attrs_field(init=False, factory=dict)
 
     def to_dict(self) -> dict[str, Any]:
@@ -79,16 +88,36 @@ def to_dict(self) -> dict[str, Any]:
         else:
             response_schema = self.response_schema
 
-        llm: str | Unset = UNSET
-        if not isinstance(self.llm, Unset):
+        llm: None | str | Unset
+        if isinstance(self.llm, Unset):
+            llm = UNSET
+        elif isinstance(self.llm, LLMEnumPublic):
             llm = self.llm.value
+        else:
+            llm = self.llm
 
-        effort_level: str | Unset = UNSET
-        if not isinstance(self.effort_level, Unset):
+        effort_level: None | str | Unset
+        if isinstance(self.effort_level, Unset):
+            effort_level = UNSET
+        elif isinstance(self.effort_level, PublicEffortLevel):
             effort_level = self.effort_level.value
+        else:
+            effort_level = self.effort_level
 
         join_with_input = self.join_with_input
 
+        iteration_budget: int | None | Unset
+        if isinstance(self.iteration_budget, Unset):
+            iteration_budget = UNSET
+        else:
+            iteration_budget = self.iteration_budget
+
+        include_research: bool | None | Unset
+        if isinstance(self.include_research, Unset):
+            include_research = UNSET
+        else:
+            include_research = self.include_research
+
         field_dict: dict[str, Any] = {}
         field_dict.update(self.additional_properties)
         field_dict.update(
@@ -107,6 +136,10 @@ def to_dict(self) -> dict[str, Any]:
             field_dict["effort_level"] = effort_level
         if join_with_input is not UNSET:
             field_dict["join_with_input"] = join_with_input
+        if iteration_budget is not UNSET:
+            field_dict["iteration_budget"] = iteration_budget
+        if include_research is not UNSET:
+            field_dict["include_research"] = include_research
 
         return field_dict
 
@@ -184,22 +217,60 @@ def _parse_response_schema(data: object) -> AgentMapOperationResponseSchemaType0
 
         response_schema = _parse_response_schema(d.pop("response_schema", UNSET))
 
-        _llm = d.pop("llm", UNSET)
-        llm: PublicLLM | Unset
-        if isinstance(_llm, Unset):
-            llm = UNSET
-        else:
-            llm = PublicLLM(_llm)
+        def _parse_llm(data: object) -> LLMEnumPublic | None | Unset:
+            if data is None:
+                return data
+            if isinstance(data, Unset):
+                return data
+            try:
+                if not isinstance(data, str):
+                    raise TypeError()
+                llm_type_0 = LLMEnumPublic(data)
 
-        _effort_level = d.pop("effort_level", UNSET)
-        effort_level: PublicEffortLevel | Unset
-        if isinstance(_effort_level, Unset):
-            effort_level = UNSET
-        else:
-            effort_level = PublicEffortLevel(_effort_level)
+                return llm_type_0
+            except (TypeError, ValueError, AttributeError, KeyError):
+                pass
+            return cast(LLMEnumPublic | None | Unset, data)
+
+        llm = _parse_llm(d.pop("llm", UNSET))
+
+        def _parse_effort_level(data: object) -> None | PublicEffortLevel | Unset:
+            if data is None:
+                return data
+            if isinstance(data, Unset):
+                return data
+            try:
+                if not isinstance(data, str):
+                    raise TypeError()
+                effort_level_type_0 = PublicEffortLevel(data)
+
+                return effort_level_type_0
+            except (TypeError, ValueError, AttributeError, KeyError):
+                pass
+            return cast(None | PublicEffortLevel | Unset, data)
+
+        effort_level = _parse_effort_level(d.pop("effort_level", UNSET))
 
         join_with_input = d.pop("join_with_input", UNSET)
 
+        def _parse_iteration_budget(data: object) -> int | None | Unset:
+            if data is None:
+                return data
+            if isinstance(data, Unset):
+                return data
+            return cast(int | None | Unset, data)
+
+        iteration_budget = _parse_iteration_budget(d.pop("iteration_budget", UNSET))
+
+        def _parse_include_research(data: object) -> bool | None | Unset:
+            if data is None:
+                return data
+            if isinstance(data, Unset):
+                return data
+            return cast(bool | None | Unset, data)
+
+        include_research = _parse_include_research(d.pop("include_research", UNSET))
+
         agent_map_operation = cls(
             input_=input_,
             task=task,
@@ -208,6 +279,8 @@ def _parse_response_schema(data: object) -> AgentMapOperationResponseSchemaType0
             llm=llm,
             effort_level=effort_level,
             join_with_input=join_with_input,
+            iteration_budget=iteration_budget,
+            include_research=include_research,
         )
 
         agent_map_operation.additional_properties = d