refactor: remove DBOS durable execution support

- Remove all DBOS integration code from agents, tools, and configuration
- Eliminate DBOSAgent wrappers and workflow management across codebase
- Remove DBOS-related environment variables and configuration options
- Delete DBOS dependency from project requirements (pyproject.toml)
- Remove DBOS-specific test infrastructure and integration tests
- Simplify agent creation by removing conditional DBOS wrapping logic
- Clean up MCP server handling by removing DBOS serialization workarounds
- Update documentation to remove DBOS configuration section
- Reduce external network dependencies (removed cloud.dbos.dev from whitelist)

Author: mpfaffenberger

README.md

@@ -133,27 +133,6 @@ These providers are automatically configured with correct OpenAI-compatible endp
 - **⚠️ Unsupported Providers** - Providers like Amazon Bedrock and Google Vertex that require special authentication are clearly marked
 - **⚠️ No Tool Calling** - Models without tool calling support show a big warning since they can't use Code Puppy's file/shell tools
 
-### Durable Execution
-
-Code Puppy now supports **[DBOS](https://github.com/dbos-inc/dbos-transact-py)** durable execution.
-
-When enabled, every agent is automatically wrapped as a `DBOSAgent`, checkpointing key interactions (including agent inputs, LLM responses, MCP calls, and tool calls) in a database for durability and recovery.
-
-You can toggle DBOS via either of these options:
-
-- CLI config (persists): `/set enable_dbos true` (or `false` to disable)
-
-
-Config takes precedence if set; otherwise the environment variable is used.
-
-### Configuration
-
-The following environment variables control DBOS behavior:
-- `DBOS_CONDUCTOR_KEY`: If set, Code Puppy connects to the [DBOS Management Console](https://console.dbos.dev/). Make sure you first register an app named `dbos-code-puppy` on the console to generate a Conductor key. Default: `None`.
-- `DBOS_LOG_LEVEL`: Logging verbosity: `CRITICAL`, `ERROR`, `WARNING`, `INFO`, or `DEBUG`. Default: `ERROR`.
-- `DBOS_SYSTEM_DATABASE_URL`: Database URL used by DBOS. Can point to a local SQLite file or a Postgres instance. Example: `postgresql://postgres:dbos@localhost:5432/postgres`. Default: `dbos_store.sqlite` file in the config directory.
-- `DBOS_APP_VERSION`: If set, Code Puppy uses it as the [DBOS application version](https://docs.dbos.dev/architecture#application-and-workflow-versions) and automatically tries to recover pending workflows for this version. Default: Code Puppy version + Unix timestamp in millisecond (disable automatic recovery).
-
 ### Custom Commands
 Create markdown files in `.claude/commands/`, `.github/prompts/`, or `.agents/commands/` to define custom slash commands. The filename becomes the command name and the content runs as a prompt.
 

code_puppy/agents/base_agent.py

@@ -24,7 +24,6 @@
 import mcp
 import pydantic
 import pydantic_ai.models
-from dbos import DBOS, SetWorkflowID
 from pydantic_ai import Agent as PydanticAgent
 from pydantic_ai import (
     BinaryContent,
@@ -35,7 +34,6 @@
     UsageLimitExceeded,
     UsageLimits,
 )
-from pydantic_ai.durable_exec.dbos import DBOSAgent
 from pydantic_ai.messages import (
     ModelMessage,
     ModelRequest,
@@ -56,7 +54,6 @@
     get_global_model_name,
     get_message_limit,
     get_protected_token_count,
-    get_use_dbos,
     get_value,
 )
 from code_puppy.error_logging import log_error
@@ -1212,56 +1209,25 @@ def reload_code_generation_agent(self, message_group: Optional[str] = None):
 
         self._last_model_name = resolved_model_name
         # expose for run_with_mcp
-        # Wrap it with DBOS, but handle MCP servers separately to avoid serialization issues
         global _reload_count
         _reload_count += 1
-        if get_use_dbos():
-            # Don't pass MCP servers to the agent constructor when using DBOS
-            # This prevents the "cannot pickle async_generator object" error
-            # MCP servers will be handled separately in run_with_mcp
-            agent_without_mcp = PydanticAgent(
-                model=model,
-                instructions=instructions,
-                output_type=str,
-                retries=3,
-                toolsets=[],  # Don't include MCP servers here
-                history_processors=[self.message_history_accumulator],
-                model_settings=model_settings,
-            )
-
-            # Register regular tools (non-MCP) on the new agent
-            agent_tools = self.get_available_tools()
-            register_tools_for_agent(agent_without_mcp, agent_tools)
-
-            # Wrap with DBOS
-            dbos_agent = DBOSAgent(
-                agent_without_mcp, name=f"{self.name}-{_reload_count}"
-            )
-            self.pydantic_agent = dbos_agent
-            self._code_generation_agent = dbos_agent
+        # Include filtered MCP servers in the agent
+        p_agent = PydanticAgent(
+            model=model,
+            instructions=instructions,
+            output_type=str,
+            retries=3,
+            toolsets=filtered_mcp_servers if filtered_mcp_servers else [],
+            history_processors=[self.message_history_accumulator],
+            model_settings=model_settings,
+        )
+        # Register regular tools on the agent
+        agent_tools = self.get_available_tools()
+        register_tools_for_agent(p_agent, agent_tools)
 
-            # Store filtered MCP servers separately for runtime use
-            self._mcp_servers = filtered_mcp_servers
-        else:
-            # Normal path without DBOS - include filtered MCP servers in the agent
-            # Re-create agent with filtered MCP servers
-            p_agent = PydanticAgent(
-                model=model,
-                instructions=instructions,
-                output_type=str,
-                retries=3,
-                toolsets=filtered_mcp_servers,
-                history_processors=[self.message_history_accumulator],
-                model_settings=model_settings,
-            )
-            # Register regular tools on the agent
-            agent_tools = self.get_available_tools()
-            register_tools_for_agent(p_agent, agent_tools)
-
-            self.pydantic_agent = p_agent
-            self._code_generation_agent = p_agent
-            self._mcp_servers = filtered_mcp_servers
-            self._mcp_servers = mcp_servers
+        self.pydantic_agent = p_agent
+        self._code_generation_agent = p_agent
+        self._mcp_servers = filtered_mcp_servers
         return self._code_generation_agent
 
     def _create_agent_with_output_type(self, output_type: Type[Any]) -> PydanticAgent:
@@ -1275,7 +1241,7 @@ def _create_agent_with_output_type(self, output_type: Type[Any]) -> PydanticAgen
             output_type: The Pydantic model or type for structured output.
 
         Returns:
-            A configured PydanticAgent (or DBOSAgent wrapper) with the custom output_type.
+            A configured PydanticAgent with the custom output_type.
         """
         from code_puppy.model_utils import prepare_prompt_for_model
         from code_puppy.tools import register_tools_for_agent
@@ -1302,38 +1268,19 @@ def _create_agent_with_output_type(self, output_type: Type[Any]) -> PydanticAgen
         global _reload_count
         _reload_count += 1
 
-        if get_use_dbos():
-            temp_agent = PydanticAgent(
-                model=model,
-                instructions=instructions,
-                output_type=output_type,
-                retries=3,
-                toolsets=[],
-                history_processors=[self.message_history_accumulator],
-                model_settings=model_settings,
-            )
-            agent_tools = self.get_available_tools()
-            register_tools_for_agent(temp_agent, agent_tools)
-            dbos_agent = DBOSAgent(
-                temp_agent, name=f"{self.name}-structured-{_reload_count}"
-            )
-            return dbos_agent
-        else:
-            temp_agent = PydanticAgent(
-                model=model,
-                instructions=instructions,
-                output_type=output_type,
-                retries=3,
-                toolsets=mcp_servers,
-                history_processors=[self.message_history_accumulator],
-                model_settings=model_settings,
-            )
-            agent_tools = self.get_available_tools()
-            register_tools_for_agent(temp_agent, agent_tools)
-            return temp_agent
+        temp_agent = PydanticAgent(
+            model=model,
+            instructions=instructions,
+            output_type=output_type,
+            retries=3,
+            toolsets=mcp_servers,
+            history_processors=[self.message_history_accumulator],
+            model_settings=model_settings,
+        )
+        agent_tools = self.get_available_tools()
+        register_tools_for_agent(temp_agent, agent_tools)
+        return temp_agent
 
-    # It's okay to decorate it with DBOS.step even if not using DBOS; the decorator is a no-op in that case.
-    @DBOS.step()
     def message_history_accumulator(self, ctx: RunContext, messages: List[Any]):
         _message_history = self.get_message_history()
         message_history_hashes = set([self.hash_message(m) for m in _message_history])
@@ -1841,49 +1788,14 @@ async def run_agent_task():
 
                 usage_limits = UsageLimits(request_limit=get_message_limit())
 
-                # Handle MCP servers - add them temporarily when using DBOS
-                if (
-                    get_use_dbos()
-                    and hasattr(self, "_mcp_servers")
-                    and self._mcp_servers
-                ):
-                    # Temporarily add MCP servers to the DBOS agent using internal _toolsets
-                    original_toolsets = pydantic_agent._toolsets
-                    pydantic_agent._toolsets = original_toolsets + self._mcp_servers
-                    pydantic_agent._toolsets = original_toolsets + self._mcp_servers
-
-                    try:
-                        # Set the workflow ID for DBOS context so DBOS and Code Puppy ID match
-                        with SetWorkflowID(group_id):
-                            result_ = await pydantic_agent.run(
-                                prompt_payload,
-                                message_history=self.get_message_history(),
-                                usage_limits=usage_limits,
-                                event_stream_handler=self._event_stream_handler,
-                                **kwargs,
-                            )
-                    finally:
-                        # Always restore original toolsets
-                        pydantic_agent._toolsets = original_toolsets
-                elif get_use_dbos():
-                    # DBOS without MCP servers
-                    with SetWorkflowID(group_id):
-                        result_ = await pydantic_agent.run(
-                            prompt_payload,
-                            message_history=self.get_message_history(),
-                            usage_limits=usage_limits,
-                            event_stream_handler=self._event_stream_handler,
-                            **kwargs,
-                        )
-                else:
-                    # Non-DBOS path (MCP servers are already included)
-                    result_ = await pydantic_agent.run(
-                        prompt_payload,
-                        message_history=self.get_message_history(),
-                        usage_limits=usage_limits,
-                        event_stream_handler=self._event_stream_handler,
-                        **kwargs,
-                    )
+                # MCP servers are already included in the agent
+                result_ = await pydantic_agent.run(
+                    prompt_payload,
+                    message_history=self.get_message_history(),
+                    usage_limits=usage_limits,
+                    event_stream_handler=self._event_stream_handler,
+                    **kwargs,
+                )
                 return result_
             except* UsageLimitExceeded as ule:
                 emit_info(f"Usage limit exceeded: {str(ule)}", group_id=group_id)
@@ -1899,12 +1811,8 @@ async def run_agent_task():
                 )
             except* asyncio.exceptions.CancelledError:
                 emit_info("Cancelled")
-                if get_use_dbos():
-                    await DBOS.cancel_workflow_async(group_id)
             except* InterruptedError as ie:
                 emit_info(f"Interrupted: {str(ie)}")
-                if get_use_dbos():
-                    await DBOS.cancel_workflow_async(group_id)
             except* Exception as other_error:
                 # Filter out CancelledError and UsageLimitExceeded from the exception group - let it propagate
                 remaining_exceptions = []

code_puppy/cli_runner.py

@@ -12,11 +12,9 @@
 import asyncio
 import os
 import sys
-import time
 import traceback
 from pathlib import Path
 
-from dbos import DBOS, DBOSConfig
 from rich.console import Console
 
 from code_puppy import __version__, callbacks, plugins
@@ -26,10 +24,8 @@
 from code_puppy.config import (
     AUTOSAVE_DIR,
     COMMAND_HISTORY_FILE,
-    DBOS_DATABASE_URL,
     ensure_config_exists,
     finalize_autosave_session,
-    get_use_dbos,
     initialize_command_history_file,
     save_command_to_history,
 )
@@ -287,33 +283,6 @@ def _uvx_protective_sigint_handler(_sig, _frame):
 
     await callbacks.on_startup()
 
-    # Initialize DBOS if not disabled
-    if get_use_dbos():
-        # Append a Unix timestamp in ms to the version for uniqueness
-        dbos_app_version = os.environ.get(
-            "DBOS_APP_VERSION", f"{current_version}-{int(time.time() * 1000)}"
-        )
-        dbos_config: DBOSConfig = {
-            "name": "dbos-code-puppy",
-            "system_database_url": DBOS_DATABASE_URL,
-            "run_admin_server": False,
-            "conductor_key": os.environ.get(
-                "DBOS_CONDUCTOR_KEY"
-            ),  # Optional, if set in env, connect to conductor
-            "log_level": os.environ.get(
-                "DBOS_LOG_LEVEL", "ERROR"
-            ),  # Default to ERROR level to suppress verbose logs
-            "application_version": dbos_app_version,  # Match DBOS app version to Code Puppy version
-        }
-        try:
-            DBOS(config=dbos_config)
-            DBOS.launch()
-        except Exception as e:
-            emit_error(f"Error initializing DBOS: {e}")
-            sys.exit(1)
-    else:
-        pass
-
     global shutdown_flag
     shutdown_flag = False
     try:
@@ -338,8 +307,6 @@ def _uvx_protective_sigint_handler(_sig, _frame):
         if bus_renderer:
             bus_renderer.stop()
         await callbacks.on_shutdown()
-        if get_use_dbos():
-            DBOS.destroy()
 
 
 async def interactive_mode(message_renderer, initial_command: str = None) -> None:
@@ -907,8 +874,6 @@ def main_entry():
     except KeyboardInterrupt:
         # Note: Using sys.stderr for crash output - messaging system may not be available
         sys.stderr.write(traceback.format_exc())
-        if get_use_dbos():
-            DBOS.destroy()
         return 0
     finally:
         # Reset terminal on Unix-like systems (not Windows)

code_puppy/command_line/config_commands.py

@@ -43,7 +43,6 @@ def handle_show_command(command: str) -> bool:
         get_protected_token_count,
         get_puppy_name,
         get_temperature,
-        get_use_dbos,
         get_yolo_mode,
     )
     from code_puppy.keymap import get_cancel_agent_display_name
@@ -72,7 +71,6 @@ def handle_show_command(command: str) -> bool:
 [bold]default_agent:[/bold]        [cyan]{default_agent}[/cyan]
 [bold]model:[/bold]                 [green]{model}[/green]
 [bold]YOLO_MODE:[/bold]             {"[red]ON[/red]" if yolo_mode else "[yellow]off[/yellow]"}
-[bold]DBOS:[/bold]                  {"[green]enabled[/green]" if get_use_dbos() else "[yellow]disabled[/yellow]"} (toggle: /set enable_dbos true|false)
 [bold]auto_save_session:[/bold]     {"[green]enabled[/green]" if auto_save else "[yellow]disabled[/yellow]"}
 [bold]protected_tokens:[/bold]      [cyan]{protected_tokens:,}[/cyan] recent tokens preserved
 [bold]compaction_threshold:[/bold]     [cyan]{compaction_threshold:.1%}[/cyan] context usage triggers compaction
@@ -213,14 +211,6 @@ def handle_set_command(command: str) -> bool:
         )
         return True
     if key:
-        # Check if we're toggling DBOS enablement
-        if key == "enable_dbos":
-            emit_info(
-                Text.from_markup(
-                    "[yellow]⚠️ DBOS configuration changed. Please restart Code Puppy for this change to take effect.[/yellow]"
-                )
-            )
-
         # Validate cancel_agent_key before setting
         if key == "cancel_agent_key":
             from code_puppy.keymap import VALID_CANCEL_KEYS