Support multi-agent

agents/agents.py

@@ -92,6 +92,7 @@ class ActionStep:
     final_answer: Any = None
     error: AgentError | None = None
     step_duration: float | None = None
+    llm_output: str | None = None
 
 @dataclass
 class PlanningStep:
@@ -440,23 +441,20 @@ class ReactAgent(BaseAgent):
         else:
             self.logs.append(TaskStep(task=task))
 
-        with console.status(
+        if oneshot:
-            "Agent is running...", spinner="aesthetic"
+            step_start_time = time.time()
-        ):
+            step_log = ActionStep(start_time=step_start_time)
-            if oneshot:
+            step_log.step_end_time = time.time()
-                step_start_time = time.time()
+            step_log.step_duration = step_log.step_end_time - step_start_time
-                step_log = ActionStep(start_time=step_start_time)
-                step_log.step_end_time = time.time()
-                step_log.step_duration = step_log.step_end_time - step_start_time
 
-                # Run the agent's step
+            # Run the agent's step
-                result = self.step(step_log)
+            result = self.step(step_log)
-                return result
+            return result
 
-            if stream:
+        if stream:
-                return self.stream_run(task)
+            return self.stream_run(task)
-            else:
+        else:
-                return self.direct_run(task)
+            return self.direct_run(task)
 
     def stream_run(self, task: str):
         """
@@ -468,6 +466,9 @@ class ReactAgent(BaseAgent):
             step_start_time = time.time()
             step_log = ActionStep(iteration=iteration, start_time=step_start_time)
             try:
+                if self.planning_interval is not None and iteration % self.planning_interval == 0:
+                    self.planning_step(task, is_first_step=(iteration == 0), iteration=iteration)
+                console.rule("[bold]New step")
                 self.step(step_log)
                 if step_log.final_answer is not None:
                     final_answer = step_log.final_answer
@@ -484,7 +485,6 @@ class ReactAgent(BaseAgent):
 
         if final_answer is None and iteration == self.max_iterations:
             error_message = "Reached max iterations."
-            console.print(f"[bold red]{error_message}")
             final_step_log = ActionStep(error=AgentMaxIterationsError(error_message))
             self.logs.append(final_step_log)
             final_answer = self.provide_final_answer(task)
@@ -509,6 +509,7 @@ class ReactAgent(BaseAgent):
             try:
                 if self.planning_interval is not None and iteration % self.planning_interval == 0:
                     self.planning_step(task, is_first_step=(iteration == 0), iteration=iteration)
+                console.rule("[bold]New step")
                 self.step(step_log)
                 if step_log.final_answer is not None:
                     final_answer = step_log.final_answer
@@ -527,7 +528,6 @@ class ReactAgent(BaseAgent):
             error_message = "Reached max iterations."
             final_step_log = ActionStep(error=AgentMaxIterationsError(error_message))
             self.logs.append(final_step_log)
-            console.print(f"[bold red]{error_message}")
             final_answer = self.provide_final_answer(task)
             final_step_log.final_answer = final_answer
             final_step_log.step_duration = 0
@@ -677,7 +677,6 @@ class JsonAgent(ReactAgent):
         agent_memory = self.write_inner_memory_from_logs()
 
         self.prompt = agent_memory
-        console.rule("New step")
 
         # Add new step in logs
         log_entry.agent_memory = agent_memory.copy()
@@ -692,12 +691,14 @@ class JsonAgent(ReactAgent):
             llm_output = self.llm_engine(
                 self.prompt, stop_sequences=["<end_action>", "Observation:"], **additional_args
             )
+            log_entry.llm_output = llm_output
         except Exception as e:
             raise AgentGenerationError(f"Error in generating llm output: {e}.")
-        console.rule("Output message of the LLM")
-        console.print(llm_output)
-        log_entry.llm_output = llm_output
 
+        if self.verbose:
+            console.rule("[italic]Output message of the LLM:")
+            console.print(llm_output)
+        
         # Parse
         rationale, action = self.extract_action(llm_output=llm_output, split_token="Action:")
 
@@ -796,7 +797,6 @@ class CodeAgent(ReactAgent):
         agent_memory = self.write_inner_memory_from_logs()
 
         self.prompt = agent_memory.copy()
-        console.rule("New step")
 
         # Add new step in logs
         log_entry.agent_memory = agent_memory.copy()
@@ -811,13 +811,13 @@ class CodeAgent(ReactAgent):
             llm_output = self.llm_engine(
                 self.prompt, stop_sequences=["<end_action>", "Observation:"], **additional_args
             )
+            log_entry.llm_output = llm_output
         except Exception as e:
             raise AgentGenerationError(f"Error in generating llm output: {e}.")
 
         if self.verbose:
             console.rule("[italic]Output message of the LLM:")
             console.print(Syntax(llm_output, lexer='markdown', background_color='default'))
-        log_entry.llm_output = llm_output
 
         # Parse
         try:

agents/default_tools.py

@@ -185,3 +185,13 @@ class FinalAnswerTool(Tool):
 
     def forward(self, answer):
         return answer
+
+class UserInputTool(Tool):
+    name = "user_input"
+    description = "Asks for user's input on a specific question"
+    inputs = {"question": {"type": "string", "description": "The question to ask the user"}}
+    output_type = "string"
+
+    def forward(self, question):
+        user_input = input(f"{question} => ")
+        return user_input

examples/orchestrator.py

@@ -0,0 +1,73 @@
+import re
+import requests
+from markdownify import markdownify as md
+from requests.exceptions import RequestException
+
+from agents import (
+    tool,
+    CodeAgent,
+    JsonAgent,
+    HfApiEngine,
+    ManagedAgent,
+)
+from agents.default_tools import UserInputTool
+from agents.search import DuckDuckGoSearchTool
+from agents.utils import console
+
+model = "Qwen/Qwen2.5-72B-Instruct"
+
+@tool
+def visit_webpage(url: str) -> str:
+    """Visits a webpage at the given URL and returns its content as a markdown string.
+
+    Args:
+        url: The URL of the webpage to visit.
+
+    Returns:
+        The content of the webpage converted to Markdown, or an error message if the request fails.
+    """
+    try:
+        # Send a GET request to the URL
+        response = requests.get(url)
+        response.raise_for_status()  # Raise an exception for bad status codes
+
+        # Convert the HTML content to Markdown
+        markdown_content = md(response.text).strip()
+
+        # Remove multiple line breaks
+        markdown_content = re.sub(r"\n{3,}", "\n\n", markdown_content)
+
+        return markdown_content
+
+    except RequestException as e:
+        return f"Error fetching the webpage: {str(e)}"
+    except Exception as e:
+        return f"An unexpected error occurred: {str(e)}"
+
+
+llm_engine = HfApiEngine(model)
+
+web_agent = JsonAgent(
+    tools=[DuckDuckGoSearchTool(), visit_webpage],
+    llm_engine=llm_engine,
+    max_iterations=10,
+)
+
+managed_web_agent = ManagedAgent(
+    agent=web_agent,
+    name="search",
+    description="Runs web searches for you. Give it your query as an argument.",
+)
+
+manager_agent = CodeAgent(
+    tools=[UserInputTool()],
+    llm_engine=llm_engine,
+    managed_agents=[managed_web_agent],
+    additional_authorized_imports=["time", "datetime"],
+)
+
+with console.status(
+    "Agent is running...", spinner="aesthetic"
+):
+    manager_agent.run("""How many years ago was Stripe founded?
+    You should ask for user input on wether the answer is correct before returning your final answer.""")

poetry.lock

@@ -1,5 +1,26 @@
 # This file is automatically @generated by Poetry 1.8.2 and should not be changed by hand.
 
+[[package]]
+name = "beautifulsoup4"
+version = "4.12.3"
+description = "Screen-scraping library"
+optional = false
+python-versions = ">=3.6.0"
+files = [
+    {file = "beautifulsoup4-4.12.3-py3-none-any.whl", hash = "sha256:b80878c9f40111313e55da8ba20bdba06d8fa3969fc68304167741bbf9e082ed"},
+    {file = "beautifulsoup4-4.12.3.tar.gz", hash = "sha256:74e3d1928edc070d21748185c46e3fb33490f22f52a3addee9aee0f4f7781051"},
+]
+
+[package.dependencies]
+soupsieve = ">1.2"
+
+[package.extras]
+cchardet = ["cchardet"]
+chardet = ["chardet"]
+charset-normalizer = ["charset-normalizer"]
+html5lib = ["html5lib"]
+lxml = ["lxml"]
+
 [[package]]
 name = "certifi"
 version = "2024.8.30"
@@ -338,6 +359,21 @@ profiling = ["gprof2dot"]
 rtd = ["jupyter_sphinx", "mdit-py-plugins", "myst-parser", "pyyaml", "sphinx", "sphinx-copybutton", "sphinx-design", "sphinx_book_theme"]
 testing = ["coverage", "pytest", "pytest-cov", "pytest-regressions"]
 
+[[package]]
+name = "markdownify"
+version = "0.14.1"
+description = "Convert HTML to markdown."
+optional = false
+python-versions = "*"
+files = [
+    {file = "markdownify-0.14.1-py3-none-any.whl", hash = "sha256:4c46a6c0c12c6005ddcd49b45a5a890398b002ef51380cd319db62df5e09bc2a"},
+    {file = "markdownify-0.14.1.tar.gz", hash = "sha256:a62a7a216947ed0b8dafb95b99b2ef4a0edd1e18d5653c656f68f03db2bfb2f1"},
+]
+
+[package.dependencies]
+beautifulsoup4 = ">=4.9,<5"
+six = ">=1.15,<2"
+
 [[package]]
 name = "markupsafe"
 version = "3.0.2"
@@ -1096,6 +1132,17 @@ files = [
     {file = "six-1.17.0.tar.gz", hash = "sha256:ff70335d468e7eb6ec65b95b99d3a2836546063f63acc5171de367e834932a81"},
 ]
 
+[[package]]
+name = "soupsieve"
+version = "2.6"
+description = "A modern CSS selector implementation for Beautiful Soup."
+optional = false
+python-versions = ">=3.8"
+files = [
+    {file = "soupsieve-2.6-py3-none-any.whl", hash = "sha256:e72c4ff06e4fb6e4b5a9f0f55fe6e81514581fca1515028625d0f299c602ccc9"},
+    {file = "soupsieve-2.6.tar.gz", hash = "sha256:e2e68417777af359ec65daac1057404a3c8a5455bb8abc36f1a9866ab1a51abb"},
+]
+
 [[package]]
 name = "tokenizers"
 version = "0.21.0"
@@ -1301,4 +1348,4 @@ zstd = ["zstandard (>=0.18.0)"]
 [metadata]
 lock-version = "2.0"
 python-versions = ">=3.10,<3.13"
-content-hash = "6c3841968936d66bf70e11c6c8e0a16fec6c2f4d88d79cd8ac5a412225e7cf56"
+content-hash = "3a0896faf882952a0d780efcc862017989612fcb421a6ee01e4eec0ba6c0f638"

pyproject.toml

@@ -67,6 +67,7 @@ pandas = "^2.2.3"
 jinja2 = "^3.1.4"
 pillow = "^11.0.0"
 llama-cpp-python = "^0.3.4"
+markdownify = "^0.14.1"
 
 
 [build-system]