feat: support opaque_body from plugin LLM calls #31566

BeautyyuYanli · 2026-01-26T17:15:51Z

Summary

This is a part of langgenius/dify-official-plugins#2492

Adds end-to-end support for preserving assistant opaque_body produced by plugin LLM calls.

Adds message-level and content-level opaque_body fields in model runtime prompt/message entities.
Parses opaque_body from plugin LLM invocations (including streaming tool-call parsing).
Captures assistant opaque_body during app run streaming, persists it into message_metadata as assistant_opaque_body, and restores it when rebuilding memory prompt history.

Testing

Added unit tests: api/tests/unit_tests/core/model_runtime/__base/test_llm_invoke_opaque_body.py

…om plugin LLM call (and fix the tool call parsing in streaming mode)

gemini-code-assist · 2026-01-26T17:16:06Z

Summary of Changes

Hello @BeautyyuYanli, I'm Gemini Code Assist¹! I'm currently reviewing this pull request and will post my feedback shortly. In the meantime, here's a summary to help you and other reviewers quickly get up to speed!

This pull request significantly enhances the system's capability to handle and retain custom, unstructured data associated with LLM responses from plugins. By integrating opaque_body fields into core message entities and ensuring its persistence and retrieval across various stages of the application's lifecycle, the change allows for more flexible and powerful interactions with LLMs that provide additional metadata.

Highlights

Enhanced Opaque Body Support: Introduced end-to-end support for preserving opaque_body produced by plugin LLM calls, ensuring this custom data is maintained throughout the system.
Model Runtime Integration: Added opaque_body fields at both message-level and content-level within model runtime prompt and message entities to accommodate this new data.
Plugin LLM Invocation Parsing: Implemented logic to parse opaque_body from plugin LLM invocations, including robust handling for streaming tool-call parsing.
Persistence and Memory Reconstruction: Ensured that the assistant's opaque_body is captured during app run streaming, persisted into message_metadata as assistant_opaque_body, and correctly restored when rebuilding memory prompt history.
Comprehensive Testing: New unit tests have been added to validate the correct preservation of opaque_body in both non-streaming and streaming LLM invocation scenarios.

🧠 New Feature in Public Preview: You can now enable Memory to help Gemini Code Assist learn from your team's feedback. This makes future code reviews more consistent and personalized to your project's style. Click here to enable Memory in your admin console.

Using Gemini Code Assist

The full guide for Gemini Code Assist can be found on our documentation page, here are some quick tips.

Invoking Gemini

You can request assistance from Gemini at any point by creating a comment using either /gemini <command> or @gemini-code-assist <command>. Below is a summary of the supported commands on the current page.

Feature	Command	Description
Code Review	`/gemini review`	Performs a code review for the current pull request in its current state.
Pull Request Summary	`/gemini summary`	Provides a summary of the current pull request in its current state.
Comment	@gemini-code-assist	Responds in comments when explicitly tagged, both in pull request comments and review comments.
Help	`/gemini help`	Displays a list of available commands.

Customization

To customize Gemini Code Assist for GitHub experience, repository maintainers can create a configuration file and/or provide a custom code review style guide (such as PEP-8 for Python) by creating and adding files to a .gemini/ folder in the base of the repository. Detailed instructions can be found here.

Limitations & Feedback

Gemini Code Assist may make mistakes. Please leave feedback on any instances where its feedback is incorrect or counter productive. You can react with 👍 and 👎 on @gemini-code-assist comments. If you're interested in giving your feedback about your experience with Gemini Code Assist for Github and other Google products, sign up here.

You can also get AI-powered code generation, chat, as well as code reviews directly in the IDE at no cost with the Gemini Code Assist IDE Extension.

Review the Privacy Notices, Generative AI Prohibited Use Policy, Terms of Service, and learn how to configure Gemini Code Assist in GitHub here. Gemini can make mistakes, so double check it and use code with caution. ↩

gemini-code-assist

Code Review

The pull request introduces end-to-end support for preserving the assistant's opaque_body from plugin LLM calls. This includes adding opaque_body fields to message entities, parsing it during LLM invocations (both streaming and non-streaming), persisting it in message metadata, and restoring it when rebuilding memory prompt history. The changes are well-implemented and supported by new unit tests.

Copilot

Pull request overview

Adds runtime + app-layer plumbing to preserve and persist an assistant opaque_body returned from plugin-based LLM calls, including streaming flows, so it can be restored into prompt history later.

Changes:

Extend model-runtime message entities to carry opaque_body at both message and content levels.
Preserve assistant opaque_body when aggregating plugin LLM chunks (streaming and non-streaming) and expose it in after-invoke callbacks.
Capture assistant opaque_body during app streaming, persist it into message_metadata as assistant_opaque_body, and restore it in TokenBufferMemory.

Reviewed changes

Copilot reviewed 6 out of 7 changed files in this pull request and generated 7 comments.

Show a summary per file

File	Description
api/tests/unit_tests/core/model_runtime/__base/test_llm_invoke_opaque_body.py	Adds unit coverage ensuring `opaque_body` survives plugin LLM invoke (stream/non-stream).
api/core/model_runtime/model_providers/__base/large_language_model.py	Aggregates `opaque_body` (and tool calls) across chunks into final `LLMResult` / after-callback result.
api/core/model_runtime/entities/message_entities.py	Introduces `opaque_body` fields on prompt content and assistant messages.
api/core/memory/token_buffer_memory.py	Restores persisted `assistant_opaque_body` from `message_metadata` into `AssistantPromptMessage`.
api/core/app/task_pipeline/easy_ui_based_generate_task_pipeline.py	Persists assistant `opaque_body` into `message_metadata` under `assistant_opaque_body`.
api/core/app/apps/base_app_runner.py	Captures `opaque_body` during streaming and includes it in the emitted final `LLMResult`.
agent-notes/.gitkeep	No functional change (repo housekeeping).

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

api/core/app/task_pipeline/easy_ui_based_generate_task_pipeline.py

api/core/memory/token_buffer_memory.py

api/core/app/task_pipeline/easy_ui_based_generate_task_pipeline.py

Copilot · 2026-01-26T17:21:57Z

api/tests/unit_tests/core/model_runtime/__base/test_llm_invoke_opaque_body.py

+    def __init__(self) -> None:
+        self.after_result = None
+
+    def on_before_invoke(self, **kwargs: Any) -> None:  # noqa: ANN401


This method requires 1 positional argument, whereas overridden Callback.on_before_invoke requires at least 6.

Suggested change

def on_before_invoke(self, **kwargs: Any) -> None: # noqa: ANN401

def on_before_invoke( # noqa: ANN401

self,

arg1: Any,

arg2: Any,

arg3: Any,

arg4: Any,

arg5: Any,

arg6: Any,

*args: Any,

**kwargs: Any,

) -> None:

Copilot · 2026-01-26T17:21:58Z

api/tests/unit_tests/core/model_runtime/__base/test_llm_invoke_opaque_body.py

+    def on_before_invoke(self, **kwargs: Any) -> None:  # noqa: ANN401
+        return None
+
+    def on_new_chunk(self, **kwargs: Any) -> None:  # noqa: ANN401


This method requires 1 positional argument, whereas overridden Callback.on_new_chunk requires at least 7.

Suggested change

def on_new_chunk(self, **kwargs: Any) -> None: # noqa: ANN401

def on_new_chunk(self, *args: Any, **kwargs: Any) -> None: # noqa: ANN401

Copilot · 2026-01-26T17:21:58Z

api/tests/unit_tests/core/model_runtime/__base/test_llm_invoke_opaque_body.py

+    def on_new_chunk(self, **kwargs: Any) -> None:  # noqa: ANN401
+        return None
+
+    def on_after_invoke(self, result: LLMResult, **kwargs: Any) -> None:  # noqa: ANN401


This method requires 2 positional arguments, whereas overridden Callback.on_after_invoke requires at least 7.

Suggested change

def on_after_invoke(self, result: LLMResult, **kwargs: Any) -> None: # noqa: ANN401

def on_after_invoke(

self,

tenant_id: str | None = None, # type: ignore[unused-argument]

model_provider_instance_id: str | None = None, # type: ignore[unused-argument]

model_instance_id: str | None = None, # type: ignore[unused-argument]

model: str | None = None, # type: ignore[unused-argument]

credentials: dict[str, Any] | None = None, # type: ignore[unused-argument]

prompt_messages: list[PromptMessage] | None = None, # type: ignore[unused-argument]

result: LLMResult | None = None,

**kwargs: Any, # noqa: ANN401

) -> None:

Copilot · 2026-01-26T17:21:58Z

api/tests/unit_tests/core/model_runtime/__base/test_llm_invoke_opaque_body.py

+    def on_after_invoke(self, result: LLMResult, **kwargs: Any) -> None:  # noqa: ANN401
+        self.after_result = result
+
+    def on_invoke_error(self, **kwargs: Any) -> None:  # noqa: ANN401


This method requires 1 positional argument, whereas overridden Callback.on_invoke_error requires at least 7.

Suggested change

def on_invoke_error(self, **kwargs: Any) -> None: # noqa: ANN401

def on_invoke_error(self, *args: Any, **kwargs: Any) -> None: # noqa: ANN401

BeautyyuYanli added 4 commits January 25, 2026 22:35

add message level and content level opaque_body

5c51e0a

make the model_runtime support reading and parsing the opaque_body fr…

289b592

…om plugin LLM call (and fix the tool call parsing in streaming mode)

add the agent-note for message_entities

2b787bf

remove the agent notes

688b2cf

BeautyyuYanli requested review from QuantumGhost and laipz8200 as code owners January 26, 2026 17:15

Copilot AI review requested due to automatic review settings January 26, 2026 17:15

BeautyyuYanli requested review from Yeuoly and crazywoola as code owners January 26, 2026 17:15

dosubot bot added the size:L This PR changes 100-499 lines, ignoring generated files. label Jan 26, 2026

dosubot bot added the 💪 enhancement New feature or request label Jan 26, 2026

Copilot started reviewing on behalf of BeautyyuYanli January 26, 2026 17:16 View session

gemini-code-assist bot reviewed Jan 26, 2026

View reviewed changes

BeautyyuYanli mentioned this pull request Jan 26, 2026

bug: When calling Google models in a multi-turn conversation, an error occurs because thought_signature is not included. langgenius/dify-official-plugins#2492

Open

6 tasks

Copilot AI reviewed Jan 26, 2026

View reviewed changes

BeautyyuYanli force-pushed the yanli/add-opaque-body branch from fb409cb to 688b2cf Compare January 26, 2026 17:22

BeautyyuYanli marked this pull request as draft January 26, 2026 17:27

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

feat: support opaque_body from plugin LLM calls #31566

feat: support opaque_body from plugin LLM calls #31566

BeautyyuYanli commented Jan 26, 2026 •

edited

Loading

Uh oh!

gemini-code-assist bot commented Jan 26, 2026

Uh oh!

gemini-code-assist bot left a comment

Uh oh!

Copilot AI left a comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Copilot AI Jan 26, 2026

Uh oh!

Copilot AI Jan 26, 2026

Uh oh!

Copilot AI Jan 26, 2026

Uh oh!

Copilot AI Jan 26, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

-    def on_before_invoke(self, **kwargs: Any) -> None:  # noqa: ANN401
+    def on_before_invoke(  # noqa: ANN401
+        self,
+        arg1: Any,
+        arg2: Any,
+        arg3: Any,
+        arg4: Any,
+        arg5: Any,
+        arg6: Any,
+        *args: Any,
+        **kwargs: Any,
+    ) -> None:

	def on_new_chunk(self, **kwargs: Any) -> None: # noqa: ANN401
	def on_new_chunk(self, args: Any, *kwargs: Any) -> None: # noqa: ANN401

-    def on_after_invoke(self, result: LLMResult, **kwargs: Any) -> None:  # noqa: ANN401
+    def on_after_invoke(
+        self,
+        tenant_id: str | None = None,  # type: ignore[unused-argument]
+        model_provider_instance_id: str | None = None,  # type: ignore[unused-argument]
+        model_instance_id: str | None = None,  # type: ignore[unused-argument]
+        model: str | None = None,  # type: ignore[unused-argument]
+        credentials: dict[str, Any] | None = None,  # type: ignore[unused-argument]
+        prompt_messages: list[PromptMessage] | None = None,  # type: ignore[unused-argument]
+        result: LLMResult | None = None,
+        **kwargs: Any,  # noqa: ANN401
+    ) -> None:

	def on_invoke_error(self, **kwargs: Any) -> None: # noqa: ANN401
	def on_invoke_error(self, args: Any, *kwargs: Any) -> None: # noqa: ANN401

feat: support opaque_body from plugin LLM calls #31566

Are you sure you want to change the base?

feat: support opaque_body from plugin LLM calls #31566

Conversation

BeautyyuYanli commented Jan 26, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Summary

Testing

Uh oh!

gemini-code-assist bot commented Jan 26, 2026

Summary of Changes

Highlights

Footnotes

Uh oh!

gemini-code-assist bot left a comment

Choose a reason for hiding this comment

Code Review

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Reviewed changes

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Copilot AI Jan 26, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Jan 26, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Jan 26, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Jan 26, 2026

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

BeautyyuYanli commented Jan 26, 2026 •

edited

Loading