diff options
Diffstat (limited to 'autogpts/autogpt/autogpt/agents/base.py')
| -rw-r--r-- | autogpts/autogpt/autogpt/agents/base.py | 58 |
1 files changed, 30 insertions, 28 deletions
diff --git a/autogpts/autogpt/autogpt/agents/base.py b/autogpts/autogpt/autogpt/agents/base.py index 1dd9ac493..846427ae7 100644 --- a/autogpts/autogpt/autogpt/agents/base.py +++ b/autogpts/autogpt/autogpt/agents/base.py @@ -12,6 +12,7 @@ if TYPE_CHECKING: from autogpt.config import Config from autogpt.core.prompting.base import PromptStrategy from autogpt.core.resource.model_providers.schema import ( + AssistantChatMessage, ChatModelInfo, ChatModelProvider, ChatModelResponse, @@ -124,7 +125,7 @@ class BaseAgentConfiguration(SystemConfiguration): class BaseAgentSettings(SystemSettings): - agent_id: Optional[str] = None + agent_id: str = "" agent_data_dir: Optional[Path] = None ai_profile: AIProfile = Field(default_factory=lambda: AIProfile(ai_name="AutoGPT")) @@ -230,10 +231,7 @@ class BaseAgent(Configurable[BaseAgentSettings], ABC): return self.config.send_token_limit or self.llm.max_tokens * 3 // 4 async def propose_action(self) -> ThoughtProcessOutput: - """Runs the agent for one cycle. - - Params: - instruction: The instruction to put at the end of the prompt. + """Proposes the next action to execute, based on the task and current state. Returns: The command name and arguments, if any, and the agent's thoughts. @@ -250,7 +248,7 @@ class BaseAgent(Configurable[BaseAgentSettings], ABC): prompt = self.on_before_think(prompt, scratchpad=self._prompt_scratchpad) logger.debug(f"Executing prompt:\n{dump_prompt(prompt)}") - raw_response = await self.llm_provider.create_chat_completion( + response = await self.llm_provider.create_chat_completion( prompt.messages, functions=get_openai_command_specs( self.command_registry.list_available_commands(self) @@ -259,11 +257,16 @@ class BaseAgent(Configurable[BaseAgentSettings], ABC): if self.config.use_functions_api else [], model_name=self.llm.name, + completion_parser=lambda r: self.parse_and_process_response( + r, + prompt, + scratchpad=self._prompt_scratchpad, + ), ) self.config.cycle_count += 1 return self.on_response( - llm_response=raw_response, + llm_response=response, prompt=prompt, scratchpad=self._prompt_scratchpad, ) @@ -283,7 +286,7 @@ class BaseAgent(Configurable[BaseAgentSettings], ABC): user_input: The user's input, if any. Returns: - The results of the command. + ActionResult: An object representing the result(s) of the command. """ ... @@ -294,13 +297,13 @@ class BaseAgent(Configurable[BaseAgentSettings], ABC): extra_messages: Optional[list[ChatMessage]] = None, **extras, ) -> ChatPrompt: - """Constructs and returns a prompt with the following structure: - 1. System prompt - 2. Message history of the agent, truncated & prepended with running summary as needed - 3. `cycle_instruction` + """Constructs a prompt using `self.prompt_strategy`. Params: - cycle_instruction: The final instruction for a thinking cycle + scratchpad: An object for plugins to write additional prompt elements to. + (E.g. commands, constraints, best practices) + extra_commands: Additional commands that the agent has access to. + extra_messages: Additional messages to include in the prompt. """ if not extra_commands: extra_commands = [] @@ -349,7 +352,9 @@ class BaseAgent(Configurable[BaseAgentSettings], ABC): output to the prompt. Params: - instruction: The instruction for the current cycle, also used in constructing the prompt + prompt: The prompt that is about to be executed. + scratchpad: An object for plugins to write additional prompt elements to. + (E.g. commands, constraints, best practices) Returns: The prompt to execute @@ -386,30 +391,26 @@ class BaseAgent(Configurable[BaseAgentSettings], ABC): ) -> ThoughtProcessOutput: """Called upon receiving a response from the chat model. - Adds the last/newest message in the prompt and the response to `history`, - and calls `self.parse_and_process_response()` to do the rest. + Calls `self.parse_and_process_response()`. Params: - llm_response: The raw response from the chat model - prompt: The prompt that was executed - instruction: The instruction for the current cycle, also used in constructing the prompt + llm_response: The raw response from the chat model. + prompt: The prompt that was executed. + scratchpad: An object containing additional prompt elements from plugins. + (E.g. commands, constraints, best practices) Returns: The parsed command name and command args, if any, and the agent thoughts. """ - return self.parse_and_process_response( - llm_response, - prompt, - scratchpad=scratchpad, - ) + return llm_response.parsed_result # TODO: update memory/context @abstractmethod def parse_and_process_response( self, - llm_response: ChatModelResponse, + llm_response: AssistantChatMessage, prompt: ChatPrompt, scratchpad: PromptScratchpad, ) -> ThoughtProcessOutput: @@ -419,9 +420,10 @@ class BaseAgent(Configurable[BaseAgentSettings], ABC): since the implementation depends on the role of the derivative Agent. Params: - llm_response: The raw response from the chat model - prompt: The prompt that was executed - instruction: The instruction for the current cycle, also used in constructing the prompt + llm_response: The raw response from the chat model. + prompt: The prompt that was executed. + scratchpad: An object containing additional prompt elements from plugins. + (E.g. commands, constraints, best practices) Returns: The parsed command name and command args, if any, and the agent thoughts. |