phoenix-oss/llama-stack-mirror
0
0
Fork 1
mirror of https://github.com/meta-llama/llama-stack.git synced 2025-06-27 18:50:41 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

feat: Add "instructions" support to responses API (#2205)

Browse source 
# What does this PR do?
Add support for "instructions" to the responses API. Instructions
provide a way to swap out system (or developer) messages in new
responses.


## Test Plan
unit tests added

Signed-off-by: Derek Higgins <derekh@redhat.com>
This commit is contained in:
Derek Higgins 2025-05-20 17:52:10 +01:00 • committed by GitHub
parent 1a770cf8ac
commit 3339844fda
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
6 changed files with 153 additions and 1 deletions
Download patch file Download diff file Expand all files Collapse all files

1
llama_stack/apis/agents/agents.py
View file

@ -596,6 +596,7 @@ class Agents(Protocol):
self,
input: str | list[OpenAIResponseInput],
model: str,
instructions: str | None = None,
previous_response_id: str | None = None,
store: bool | None = True,
stream: bool | None = False,

3
llama_stack/providers/inline/agents/meta_reference/agents.py
View file

@ -313,6 +313,7 @@ class MetaReferenceAgentsImpl(Agents):
self,
input: str | list[OpenAIResponseInput],
model: str,
instructions: str | None = None,
previous_response_id: str | None = None,
store: bool | None = True,
stream: bool | None = False,
@ -320,5 +321,5 @@ class MetaReferenceAgentsImpl(Agents):
tools: list[OpenAIResponseInputTool] | None = None,
) -> OpenAIResponseObject:
return await self.openai_responses_impl.create_openai_response(
input, model, previous_response_id, store, stream, temperature, tools
input, model, instructions, previous_response_id, store, stream, temperature, tools
)

7
llama_stack/providers/inline/agents/meta_reference/openai_responses.py
View file

@ -208,6 +208,10 @@ class OpenAIResponsesImpl:
return input
async def _prepend_instructions(self, messages, instructions):
if instructions:
messages.insert(0, OpenAISystemMessageParam(content=instructions))
async def get_openai_response(
self,
id: str,
@ -219,6 +223,7 @@ class OpenAIResponsesImpl:
self,
input: str | list[OpenAIResponseInput],
model: str,
instructions: str | None = None,
previous_response_id: str | None = None,
store: bool | None = True,
stream: bool | None = False,
@ -229,7 +234,9 @@ class OpenAIResponsesImpl:
input = await self._prepend_previous_response(input, previous_response_id)
messages = await _convert_response_input_to_chat_messages(input)
await self._prepend_instructions(messages, instructions)
chat_tools = await self._convert_response_tools_to_chat_tools(tools) if tools else None
chat_response = await self.inference_api.openai_chat_completion(
model=model,
messages=messages,