Release v1.0.81

[roeai-release-bot]

This PR updates the Python SDK for release 1.0.81.

Generated from:

• roe-main release branch 1-0-81
• roe-main commit b5e92f34e243e44a5eb5dfb0a6322544b9397ce9

[greptile-apps]

Greptile Summary

This release (v1.0.81) adds three new generated API endpoints — GET /v1/agents/models/, GET /v1/agents/types/, and POST /v1/tables/upload/ — along with their corresponding request/response model classes, and corrects a version number typo (1.0.801 → 1.0.81).

• New discovery endpoints: discovery_supported_models_list and discovery_agent_engine_types_list expose tenant-agnostic metadata about supported LLM models and agent engine types; both sync and async variants are generated correctly.
• New table upload endpoint: upload_table handles multipart CSV uploads via TableUploadRequest.to_multipart(); the model handles optional organization_id (UUID or null) and with_headers fields.
• Generator artifacts: Several generated files contain duplicate imports (from typing import cast, from ..types import UNSET, Unset) and an unused from urllib.parse import quote import — cosmetic issues that do not affect runtime behavior.

Confidence Score: 4/5

Safe to merge — all changes are generated code adding new endpoints and models with no modifications to existing functionality.

The only issues found are duplicate and unused imports produced by the code generator; they have no effect on runtime correctness. The UUID encoding inconsistency in to_multipart() is benign because httpx accepts both str and bytes for multipart fields. Core logic, schema alignment, and model (de)serialization are all correct.

The generated files under src/roe/_generated/api/ and src/roe/_generated/models/table_upload_request.py have minor generator artifacts noted above, but none require changes before merging.

Important Files Changed

Filename	Overview
openapi/openapi.yml	Adds three new endpoints: GET /v1/agents/models/, GET /v1/agents/types/, and POST /v1/tables/upload/ with corresponding request/response schemas. Definitions look well-formed and required fields are correctly declared.
pyproject.toml	Version corrected from the erroneous 1.0.801 to 1.0.81; uv.lock updated to match.
src/roe/_generated/api/discovery/discovery_agent_engine_types_list.py	New generated endpoint for GET /v1/agents/types/. Contains a duplicate from typing import cast import and an unused from urllib.parse import quote import; logic is otherwise correct.
src/roe/_generated/api/discovery/discovery_supported_models_list.py	New generated endpoint for GET /v1/agents/models/ with optional capability query param. Has duplicate UNSET import and unused quote import; capability filtering logic is correct.
src/roe/_generated/api/tables/upload_table.py	New generated endpoint for POST /v1/tables/upload/ using multipart form data. Unused quote import present; parses both 201 and 400 responses correctly.
src/roe/_generated/models/table_upload_request.py	New TableUploadRequest model with multipart serialization. Has a duplicate UNSET import and the UUID branch of to_multipart() passes a plain str where all other branches pass bytes.
src/roe/_generated/models/table_upload_response.py	New TableUploadResponse model; duplicate UNSET import is present but logic is correct.
src/roe/_generated/models/agent_engine_type_list.py	New AgentEngineTypeList model; serialization and deserialization look correct.
src/roe/_generated/models/supported_llm_model.py	New SupportedLLMModel model with all fields from the OpenAPI schema; from_dict / to_dict are complete and correct.
src/roe/_generated/models/supported_llm_model_list.py	New SupportedLLMModelList model; correctly nests SupportedLLMModel items via from_dict.
src/roe/_generated/models/agent_engine_type_list_engines_item.py	New AgentEngineTypeListEnginesItem model; fully dynamic (additionalProperties only) matching the schema definition.
src/roe/_generated/models/init.py	Correctly exports all four new model classes and adds them to all.

Sequence Diagram

sequenceDiagram
    participant User
    participant SDK
    participant API

    User->>SDK: discovery.sync(client) / asyncio(client)
    SDK->>API: GET /v1/agents/types/
    API-->>SDK: 200 AgentEngineTypeList
    SDK-->>User: AgentEngineTypeList

    User->>SDK: "discovery_supported_models_list.sync(client, capability=...)"
    SDK->>API: "GET /v1/agents/models/?capability=..."
    API-->>SDK: 200 SupportedLLMModelList
    SDK-->>User: SupportedLLMModelList

    User->>SDK: "upload_table.sync(client, body=TableUploadRequest)"
    SDK->>SDK: body.to_multipart() → multipart files
    SDK->>API: POST /v1/tables/upload/ (multipart/form-data)
    API-->>SDK: 201 TableUploadResponse
    API-->>SDK: 400 ErrorResponse
    SDK-->>User: "TableUploadResponse | ErrorResponse | None"

Loading

_{Reviews (1): Last reviewed commit: "Release v1.0.81" | Re-trigger Greptile}

[greptile-apps]

Duplicate cast import

from typing import cast appears twice in this file (once bundled inside from typing import Any, cast and again as a standalone import on line 11). While Python silently ignores duplicate imports, this is a generator artifact that adds noise and will confuse static-analysis tools. The same pattern appears in discovery_supported_models_list.py (duplicate from ...types import UNSET) and table_upload_request.py (duplicate from ..types import UNSET, Unset).

[greptile-apps]

Unused quote import

from urllib.parse import quote is imported but never referenced in this file. The same unused import also appears in discovery_supported_models_list.py and upload_table.py. These are likely scaffolding left over from the generator for endpoints that use path parameters; since none of these three endpoints have path segments that need URL-encoding, the import is dead code.

[greptile-apps]

Inconsistent encoding for UUID in to_multipart

When organization_id is a UUID the value passed to httpx is a plain str, but every other branch in this method encodes values to bytes. httpx accepts both, so this won't hard-crash, but the inconsistency is a generator bug. Adding .encode() makes the output uniform and matches the else branch directly below.

Suggested change

	files.append(("organization_id", (None, str(self.organization_id), "text/plain")))
	files.append(("organization_id", (None, str(self.organization_id).encode(), "text/plain")))

[jadenfix]

Closing per request.