aap-core 0.1.1.dev1__tar.gz → 0.1.3.dev1__tar.gz

aap_core-0.1.3.dev1/PKG-INFO

@@ -0,0 +1,25 @@
+Metadata-Version: 2.3
+Name: aap-core
+Version: 0.1.3.dev1
+Summary: Agent design pattern core framework
+Keywords: agent,ai,pattern,llm
+Author: Ly Hon Quang
+Author-email: Ly Hon Quang <lyhonquang@gmail.com>
+License: MIT
+Classifier: Programming Language :: Python :: 3.10
+Classifier: Programming Language :: Python :: 3.11
+Classifier: Programming Language :: Python :: 3.12
+Classifier: Programming Language :: Python :: 3.13
+Classifier: Programming Language :: Python :: 3.14
+Classifier: Programming Language :: Python :: 3.15
+Classifier: Framework :: Pydantic :: 2
+Classifier: Operating System :: OS Independent
+Requires-Dist: a2a-sdk>=0.3.12
+Requires-Dist: pandas>=2.0.0
+Requires-Dist: rouge-score>=0.1.2
+Requires-Dist: sacrebleu>=2.5.1
+Requires-Python: >=3.10.12
+Project-URL: Repository, https://github.com/quanghona/agent_design_pattern.git
+Description-Content-Type: text/markdown
+
+# AI Agent Pattern (AAP)

aap_core-0.1.3.dev1/README.md

@@ -0,0 +1 @@
+# AI Agent Pattern (AAP)

aap_core-0.1.3.dev1/pyproject.toml

@@ -0,0 +1,39 @@
+[project]
+name = "aap_core"
+version = "0.1.3.dev1"
+description = "Agent design pattern core framework"
+readme = "README.md"
+license = { text = "MIT" }
+keywords = ["agent", "ai", "pattern", "llm"]
+authors = [
+    { name = "Ly Hon Quang", email = "lyhonquang@gmail.com" }
+]
+requires-python = ">=3.10.12"
+dependencies = [
+    "a2a-sdk>=0.3.12",
+    "pandas>=2.0.0",
+    "rouge-score>=0.1.2",
+    "sacrebleu>=2.5.1",
+]
+classifiers = [
+    "Programming Language :: Python :: 3.10",
+    "Programming Language :: Python :: 3.11",
+    "Programming Language :: Python :: 3.12",
+    "Programming Language :: Python :: 3.13",
+    "Programming Language :: Python :: 3.14",
+    "Programming Language :: Python :: 3.15",
+    "Framework :: Pydantic :: 2",
+    "Operating System :: OS Independent",
+]
+
+[project.urls]
+Repository = "https://github.com/quanghona/agent_design_pattern.git"
+
+[dependency-groups]
+lint = [
+    "ruff==0.14.0",
+]
+
+[build-system]
+requires = ["uv_build>=0.8.14,<0.9.0"]
+build-backend = "uv_build"

{aap_core-0.1.1.dev1 → aap_core-0.1.3.dev1}/src/aap_core/chain.py

@@ -5,9 +5,9 @@ from typing import Callable, Generic, List, Tuple
 from pydantic import Field, PrivateAttr
 
 from .guardrail import BaseGuardRail, PassGuardRail
-from .prompt_enhancer import (
-    BasePromptEnhancer,
-    IdentityPromptEnhancer,
+from .prompt_augmenter import (
+    BasePromptAugmenter,
+    IdentityPromptAugmenter,
 )
 from .types import AgentMessage, BaseChain, ChainMessage, ChainResponse
 
@@ -98,7 +98,7 @@ class BaseCausalMultiTurnsChain(BaseLLMChain, Generic[ChainMessage, ChainRespons
 class TypicalLLMChain(BaseLLMChain):
     """
     A LLM chain that consists of all components in a typical response generation.
-    The chain  have 4 steps:
+    The chain has 4 steps:
     - input guardrail: validate the input message
     - prompt enhancement: add more context to the prompt or rewrite / refine the prompt
     - generate: generate the response. This steps should also handle the tool calling procedure and finalize the response
@@ -108,8 +108,8 @@ class TypicalLLMChain(BaseLLMChain):
         default=PassGuardRail(),
         description="validate the input message",
     )
-    prompt_enhancer: BasePromptEnhancer = Field(
-        default=IdentityPromptEnhancer(),
+    prompt_augmenter: BasePromptAugmenter = Field(
+        default=IdentityPromptAugmenter(),
         description="add more context to the prompt or rewrite / refine the prompt",
     )
     tools: Sequence[Callable] = Field(
@@ -128,21 +128,17 @@ class TypicalLLMChain(BaseLLMChain):
 
     def invoke(self, message: AgentMessage, **kwargs) -> AgentMessage:
         message = self.input_guardrail(message)
-        message = self.prompt_enhancer(message)
+        message = self.prompt_augmenter(message)
         message = self.generate(message, **kwargs)
         message = self.output_guardrail(message)
         return message
 
 
-class LLMPromptEnhancer(BasePromptEnhancer):
-    """A prompt enhancer that use an LLM chain to rewrite the prompt."""
+class MetaPromptAugmenter(BasePromptAugmenter):
+    """A prompt augmenter that use an LLM chain to rewrite the prompt."""
 
     chain: BaseLLMChain = Field(..., description="LLM chain that rewrite the prompt")
 
-    def __call__(self, message: AgentMessage, **kwargs) -> AgentMessage:
+    def augment(self, message: AgentMessage, **kwargs) -> AgentMessage:
         message = self.chain.invoke(message, **kwargs)
-        if message.execution_result != "success":
-            return message
-        _, message.query = message.responses[-1]
-        message.responses = []
         return message

{aap_core-0.1.1.dev1 → aap_core-0.1.3.dev1}/src/aap_core/orchestration.py

@@ -35,11 +35,6 @@ class ReflectionAgent(BaseAgent):
         ...,
         description="LLM chain that perform reflection on the result of the main task",
     )
-    task_response_key: str = Field(
-        default="context_response",
-        description="""The key of which to store the context (e.g. intermediate result) during generation.
-        Note that the key need to start with 'context_'""",
-    )
 
     @field_validator("task_response_key")
     @classmethod
@@ -61,11 +56,7 @@ class ReflectionAgent(BaseAgent):
             return message
 
         self.state = "reflecting"
-        task_response_key = self.task_response_key.replace("context_", "")
         reflect_message = message.model_copy(deep=True)
-        reflect_message.context = {}
-        _, reflect_message.context[task_response_key] = reflect_message.responses[-1]  # type: ignore
-        reflect_message.responses = []
         result_message = self.chain_reflection.invoke(reflect_message, **kwargs)
         self.state = "idle"
         result_message.origin = self.card.name
@@ -343,6 +334,7 @@ class CoordinatorAgent(BaseAgent):
         # TODO: check step dependency and parallelize steps to speed up
         summary_steps_key = self.summary_steps_key.replace("context_", "")
         for i, (step_msg, step_agent, dependencies) in enumerate(steps):
+            # TODO: deal with the dependencies
             step_msg.context = {summary_steps_key: sub_task_result}
             self.state = f"step {i}: worker {step_agent.card.name} running"
             sub_task_result.append(step_agent.execute(step_msg, **kwargs))