agentic-cli 0.4.3__tar.gz → 0.4.4__tar.gz

{agentic_cli-0.4.3 → agentic_cli-0.4.4}/CHANGELOG.md

@@ -5,6 +5,38 @@ All notable changes to this project will be documented in this file.
 The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.1.0/),
 and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
 
+## [0.4.4] - 2026-03-11
+
+### Added
+- **Session Save/Resume**: Persistent conversations across CLI exits with `/save` and `/resume`
+- **Sandbox Executor**: Stateful multi-turn Python code execution with sandboxing and default deps
+- **Context Window Management**: Native ADK and LangGraph context trimming with source-level detection
+- **Context Window Visibility**: Trim detection, status bar token display, and `/status` breakdown
+- **Dynamic Model Registry**: Replace static model lists with live API discovery from Google/Anthropic
+- **Task Progress Display**: Rich-colored task progress in thinking box, persisted across turns
+
+### Changed
+- **Architecture**: Separated workflow+tools layer from UI/CLI layer; composable settings mixins
+- **DRY/SOLID Cleanup** (PR #59, #60): Deduplicated config paths, ArXiv parsing, session persistence, artifact loading; extracted `drain_trim_events()`, `format_detail_rows()`, `format_task_checklist()`; removed dead code (`get_help()`, `FinanceResearchState`, unused settings introspection); replaced dynamic `type()` with `_SessionEvent` class; moved token-drop heuristic into `UsageTracker`
+- **MessageProcessor**: Removed dead code, unused params, cleaner state init
+- **HITL**: Removed dead Future-based machinery, simplified callback path
+- **Ripgrep**: Cache availability check with `lru_cache`
+
+### Fixed
+- Executor import validation blocking allowed submodules
+- Settings not persisting (replaced `exclude_defaults` with explicit exclusion)
+- `verbose_thinking` not persisting across restarts
+- Thread safety, path traversal, and TOCTOU race conditions in persistence
+- VectorStore crash safety: write mappings before FAISS index
+- Embeddings: update in-memory state after `ingest_document`
+- `open_document`: extension allowlist, DANGEROUS permission, audit logging
+
+### Removed
+- `FinanceResearchState` (domain-specific code in shared library)
+- Dead `get_help()`, `get_ui_section`, `is_ui_hidden` methods
+- Dead `generate_tool_summary`, `clear_context`/`unbind_context`, model constant re-exports
+- Dead `cli/settings.py` re-export shim
+
 ## [0.4.3] - 2026-02-12
 
 ### Changed

{agentic_cli-0.4.3 → agentic_cli-0.4.4}/CLAUDE.md

@@ -20,7 +20,6 @@ agentic-cli/
 │   ├── __init__.py           # Package exports, lazy imports
 │   ├── config.py             # BaseSettings (pydantic-settings)
 │   ├── constants.py          # Shared constants, truncate()
-│   ├── resolvers.py          # Model/path constants (GOOGLE_MODELS, etc.)
 │   ├── settings_persistence.py
 │   ├── logging.py
 │   ├── cli/

{agentic_cli-0.4.3 → agentic_cli-0.4.4}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: agentic-cli
-Version: 0.4.3
+Version: 0.4.4
 Summary: A framework for building domain-specific agentic CLI applications
 Project-URL: Homepage, https://github.com/shoom1/agentic-cli
 Project-URL: Repository, https://github.com/shoom1/agentic-cli
@@ -17,6 +17,8 @@ Requires-Python: >=3.12
 Requires-Dist: feedparser>=6.0.0
 Requires-Dist: google-adk[genai]>=0.4.0
 Requires-Dist: httpx>=0.27.0
+Requires-Dist: ipykernel>=6.0.0
+Requires-Dist: jupyter-client>=8.0.0
 Requires-Dist: numpy>=1.26.0
 Requires-Dist: prompt-toolkit>=3.0.0
 Requires-Dist: pydantic-settings>=2.0.0
@@ -24,7 +26,7 @@ Requires-Dist: pydantic>=2.0.0
 Requires-Dist: pypdf>=4.0.0
 Requires-Dist: rich>=13.0.0
 Requires-Dist: structlog>=24.0.0
-Requires-Dist: thinking-prompt>=0.2.3
+Requires-Dist: thinking-prompt>=0.2.5
 Provides-Extra: dev
 Requires-Dist: pytest-asyncio>=0.24.0; extra == 'dev'
 Requires-Dist: pytest-cov>=6.0.0; extra == 'dev'

{agentic_cli-0.4.3 → agentic_cli-0.4.4}/examples/research_demo/agents.py

@@ -108,7 +108,7 @@ When the user asks you to research something:
 6. For arXiv paper research, **delegate to arxiv_specialist**
 7. Execute ONE task at a time, updating the plan after each
 8. Use `web_fetch` to extract information from specific URLs found during research
-9. Use `execute_python` for data analysis, calculations, or processing
+9. Use `execute_python` for quick calculations and data validation
 10. Use `ask_clarification` when you need user input to proceed
 11. Update the plan with `save_plan` if you discover changes are needed
 12. Store learnings with `save_memory` and share them with the user
@@ -156,7 +156,8 @@ AGENT_CONFIGS = [
             # arXiv (2 tools)
             search_arxiv,
             fetch_arxiv_paper,
-            # Document store (3 tools)
+            # Document store (4 tools)
+            search_knowledge_base,
             ingest_document,
             list_documents,
             read_document,

{agentic_cli-0.4.3 → agentic_cli-0.4.4}/examples/research_demo/settings.py

@@ -2,7 +2,6 @@
 
 from pathlib import Path
 
-from pydantic import Field
 from pydantic_settings import SettingsConfigDict
 
 from agentic_cli import BaseSettings
@@ -27,10 +26,19 @@ class ResearchDemoSettings(BaseSettings):
         extra="ignore",
     )
 
-    app_name: str = Field(default="research_demo")
-    workspace_dir: Path = Field(default=Path.home() / ".research_demo")
-
     def __init__(self, **kwargs):
-        # Override verbose_thinking default to False (concise mode)
-        kwargs.setdefault("verbose_thinking", False)
+        kwargs.setdefault("app_name", "research_demo")
+        kwargs.setdefault("workspace_dir", Path.home() / ".research_demo")
         super().__init__(**kwargs)
+
+    def model_post_init(self, __context):
+        """Override verbose_thinking default without blocking JSON persistence.
+
+        kwargs.setdefault would inject at init level (highest priority),
+        overriding saved JSON values. model_post_init runs after all sources
+        are resolved, and model_fields_set tracks which fields were explicitly
+        set by any source (env, JSON, init kwargs). We only apply our custom
+        default when no source provided a value.
+        """
+        if "verbose_thinking" not in self.model_fields_set:
+            object.__setattr__(self, "verbose_thinking", False)

{agentic_cli-0.4.3 → agentic_cli-0.4.4}/examples/websearch_demo.py

@@ -22,7 +22,7 @@ import sys
 
 from agentic_cli.tools.search import (
     web_search,
-    SearchResult,
+    WebSearchResult,
     TavilyBackend,
     BraveBackend,
     SEARCH_BACKENDS,
@@ -89,20 +89,20 @@ def demo_configuration_check():
 
 
 def demo_search_result_format():
-    """Demo the SearchResult data structure."""
+    """Demo the WebSearchResult data structure."""
     print("\n" + "=" * 60)
-    print("SearchResult Data Structure")
+    print("WebSearchResult Data Structure")
     print("=" * 60)
 
     # Create sample result
-    sample = SearchResult(
+    sample = WebSearchResult(
         title="Example Search Result",
         url="https://example.com/article",
         snippet="This is a sample snippet showing what search results look like...",
         score=0.95,
     )
 
-    print("  SearchResult fields:")
+    print("  WebSearchResult fields:")
     print(f"    title:   {sample.title}")
     print(f"    url:     {sample.url}")
     print(f"    snippet: {sample.snippet[:50]}...")

{agentic_cli-0.4.3 → agentic_cli-0.4.4}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "hatchling.build"
 
 [project]
 name = "agentic-cli"
-version = "0.4.3"
+version = "0.4.4"
 description = "A framework for building domain-specific agentic CLI applications"
 readme = "README.md"
 license = "MIT"
@@ -21,7 +21,7 @@ classifiers = [
     "Programming Language :: Python :: 3.13",
 ]
 dependencies = [
-    "thinking-prompt>=0.2.3",
+    "thinking-prompt>=0.2.5",
     "google-adk[genai]>=0.4.0",
     "pydantic>=2.0.0",
     "pydantic-settings>=2.0.0",
@@ -32,6 +32,8 @@ dependencies = [
     "numpy>=1.26.0",
     "feedparser>=6.0.0",
     "pypdf>=4.0.0",
+    "jupyter_client>=8.0.0",
+    "ipykernel>=6.0.0",
 ]
 
 [project.optional-dependencies]

agentic_cli-0.4.4/requirements.txt

@@ -0,0 +1,4 @@
+# Install from pyproject.toml (single source of truth for dependencies)
+# Note: thinking-prompt is a local dependency, install separately:
+#   pip install -e /path/to/thinking-prompt
+-e .

{agentic_cli-0.4.3 → agentic_cli-0.4.4}/src/agentic_cli/__init__.py

@@ -23,7 +23,7 @@ Note: GoogleADKWorkflowManager and LangGraphWorkflowManager are lazy-loaded to a
 """
 
 from agentic_cli.cli.app import BaseCLIApp
-from agentic_cli.cli.workflow_controller import create_workflow_manager_from_settings
+from agentic_cli.workflow.factory import create_workflow_manager_from_settings
 from agentic_cli.cli.commands import Command, CommandRegistry
 from agentic_cli.workflow.config import AgentConfig
 from agentic_cli.workflow.events import WorkflowEvent, EventType
@@ -40,7 +40,7 @@ from agentic_cli.config import (
 )
 from agentic_cli.settings_persistence import SettingsPersistence
 from agentic_cli.workflow.settings import WorkflowSettingsMixin
-from agentic_cli.cli.settings import CLISettingsMixin
+from agentic_cli.settings_mixins import AppSettingsMixin, CLISettingsMixin
 
 # Heavy imports - lazy loaded on first access
 _lazy_imports = {
@@ -88,7 +88,8 @@ __all__ = [
     "reload_settings",
     # Settings Mixins (organized settings by domain)
     "WorkflowSettingsMixin",
+    "AppSettingsMixin",
     "CLISettingsMixin",
 ]
 
-__version__ = "0.4.3"
+__version__ = "0.4.4"

{agentic_cli-0.4.3 → agentic_cli-0.4.4}/src/agentic_cli/cli/__init__.py

@@ -11,7 +11,7 @@ from agentic_cli.cli.commands import (
 from agentic_cli.cli.app import BaseCLIApp
 from agentic_cli.cli.message_processor import MessageProcessor, MessageHistory, MessageType
 from agentic_cli.cli.workflow_controller import WorkflowController
-from agentic_cli.cli.settings import CLISettingsMixin
+from agentic_cli.settings_mixins import CLISettingsMixin
 
 __all__ = [
     "AppInfo",

{agentic_cli-0.4.3 → agentic_cli-0.4.4}/src/agentic_cli/cli/app.py

@@ -111,6 +111,7 @@ class BaseCLIApp:
         app_info: AppInfo,
         agent_configs: list["AgentConfig"],
         settings: BaseSettings,
+        session_id: str | None = None,
     ) -> None:
         """Initialize the CLI application.
 
@@ -118,7 +119,10 @@ class BaseCLIApp:
             app_info: Application info (name, version, welcome message)
             agent_configs: List of agent configurations for the workflow
             settings: Application settings instance
+            session_id: Optional session ID for save/resume. If provided,
+                       the session will be loaded on startup and saved on exit.
         """
+        self._session_id = session_id
         # === Configuration ===
         self._app_info = app_info
         self._settings = settings
@@ -282,6 +286,16 @@ class BaseCLIApp:
         """Get the application settings."""
         return self._settings
 
+    @property
+    def session_id(self) -> str | None:
+        """Get the persistent session ID, if any."""
+        return self._session_id
+
+    @session_id.setter
+    def session_id(self, value: str | None) -> None:
+        """Set the persistent session ID."""
+        self._session_id = value
+
     def stop(self) -> None:
         """Stop the application."""
         self.should_exit = True
@@ -305,32 +319,18 @@ class BaseCLIApp:
         needs_reinit = False
         new_model = changes.get("model")
 
-        # Settings that require special handling
-        special_settings = {"model", "thinking_effort"}
-
-        # Apply model change (requires dedicated setter + reinit)
-        if new_model:
-            try:
-                self._settings.set_model(new_model)
-                needs_reinit = True
-            except ValueError as e:
-                self.session.add_error(f"Failed to set model: {e}")
-                return
+        reinit_settings = {"model", "thinking_effort"}
 
-        # Apply thinking effort change (requires dedicated setter + reinit)
-        if "thinking_effort" in changes:
+        for key, value in changes.items():
             try:
-                self._settings.set_thinking_effort(changes["thinking_effort"])
-                needs_reinit = True
+                self._settings.update_setting(key, value)
+                if key in reinit_settings:
+                    needs_reinit = True
             except ValueError as e:
-                self.session.add_error(f"Failed to set thinking effort: {e}")
+                label = key.replace("_", " ").title()
+                self.session.add_error(f"Failed to set {label}: {e}")
                 return
 
-        # Apply all other settings directly
-        for key, value in changes.items():
-            if key not in special_settings:
-                object.__setattr__(self._settings, key, value)
-
         # Reinitialize workflow if needed
         if needs_reinit and self._workflow_controller.is_ready:
             try:
@@ -351,7 +351,9 @@ class BaseCLIApp:
             ClearCommand,
             ExitCommand,
             StatusCommand,
+            SandboxCommand,
             PapersCommand,
+            SessionsCommand,
         )
         from agentic_cli.cli.settings_command import SettingsCommand
 
@@ -359,8 +361,10 @@ class BaseCLIApp:
         self.command_registry.register(ClearCommand())
         self.command_registry.register(ExitCommand())
         self.command_registry.register(StatusCommand())
+        self.command_registry.register(SandboxCommand())
         self.command_registry.register(SettingsCommand())
         self.command_registry.register(PapersCommand())
+        self.command_registry.register(SessionsCommand())
 
     async def process_input(self, user_input: str) -> None:
         """Process user input.
@@ -448,11 +452,45 @@ class BaseCLIApp:
         except Exception as e:
             logger.error("activity_log_save_failed", error=str(e))
 
+    async def _load_session_on_startup(self) -> None:
+        """Load a saved session after workflow initialization."""
+        if not self._session_id:
+            return
+
+        if not await self._workflow_controller.ensure_initialized(self.session):
+            self.session.add_warning("Cannot load session — workflow not initialized.")
+            return
+
+        workflow = self._workflow_controller.workflow
+        loaded = await workflow.load_session(self._session_id)
+        if loaded:
+            self.session.add_success(f"Session '{self._session_id}' resumed.")
+        else:
+            self.session.add_message("system", f"New session '{self._session_id}'.")
+
+    async def _save_session_on_exit(self) -> None:
+        """Save the current session on exit."""
+        if not self._session_id:
+            return
+
+        if not self._workflow_controller.is_ready:
+            return
+
+        workflow = self._workflow_controller.workflow
+        result = await workflow.save_session(self._session_id)
+        if result.get("success"):
+            logger.info("session_saved_on_exit", session_id=self._session_id)
+        else:
+            logger.error("session_save_on_exit_failed", error=result.get("error"))
+
     async def run(self) -> None:
         """Run the main application loop."""
         logger.info("repl_starting")
 
         async with self._workflow_controller.background_init(self.session):
+            if self._session_id:
+                await self._load_session_on_startup()
+
             # Register input handler
             @self.session.on_input
             async def handle_input(text: str) -> None:
@@ -463,6 +501,10 @@ class BaseCLIApp:
             # Run the session - user sees prompt immediately!
             await self.session.run_async()
 
+        # Save persistent session on exit
+        if self._session_id:
+            await self._save_session_on_exit()
+
         # Auto-save activity log if enabled
         if self._settings.log_activity and len(self.message_history) > 0:
             await self._save_activity_log()

{agentic_cli-0.4.3 → agentic_cli-0.4.4}/src/agentic_cli/cli/builtin_commands.py

@@ -108,33 +108,98 @@ class StatusCommand(Command):
             if init_error:
                 table.add_row("Error", f"[red]{init_error}[/red]")
 
+        # Persistent session info
+        sid = app.session_id
+        if sid:
+            table.add_row("Session", f"{sid} (persistent)")
+        else:
+            table.add_row("Session", "ephemeral (not saved)")
+
         # Message history stats
         table.add_row("Messages", str(len(app.message_history)))
 
         # Token usage breakdown
         tracker = getattr(app, "usage_tracker", None)
-        if tracker is not None and tracker.invocation_count > 0:
-            from agentic_cli.cli.usage_tracker import format_tokens
-
-            table.add_row("", "")  # Spacer
-            table.add_row("LLM Invocations", str(tracker.invocation_count))
-            table.add_row("Input Tokens", format_tokens(tracker.prompt_tokens))
-            table.add_row("Output Tokens", format_tokens(tracker.completion_tokens))
-            table.add_row("Total Tokens", format_tokens(tracker.total_tokens))
-            if tracker.cached_tokens > 0:
-                table.add_row("Cached Tokens", format_tokens(tracker.cached_tokens))
-            if tracker.cache_creation_tokens > 0:
-                table.add_row("Cache Creation", format_tokens(tracker.cache_creation_tokens))
-            if tracker.thinking_tokens > 0:
-                table.add_row("Thinking Tokens", format_tokens(tracker.thinking_tokens))
-            if tracker.total_latency_ms > 0:
-                avg_ms = tracker.total_latency_ms / tracker.invocation_count
-                table.add_row("Avg Latency", f"{avg_ms:.0f}ms")
+        if tracker is not None:
+            for label, value in tracker.format_detail_rows():
+                table.add_row(label, value)
 
         panel = Panel(table, title="[bold]Session Status[/bold]", border_style="cyan")
         app.session.add_rich(panel)
 
 
+class SandboxCommand(Command):
+    """Manage sandbox sessions."""
+
+    def __init__(self) -> None:
+        super().__init__(
+            name="sandbox",
+            description="List and manage sandbox sessions",
+            aliases=["sb"],
+            usage="/sandbox [reset [session_id] [--all]]",
+            examples=["/sandbox", "/sandbox reset", "/sandbox reset my_session", "/sandbox reset --all"],
+            category=CommandCategory.WORKFLOW,
+        )
+
+    async def execute(self, args: str, app: Any) -> None:
+        """List or reset sandbox sessions."""
+        # Get sandbox manager from workflow
+        try:
+            workflow = app.workflow
+            manager = getattr(workflow, "sandbox_manager", None)
+        except (RuntimeError, AttributeError):
+            manager = None
+
+        if manager is None:
+            app.session.add_warning(
+                "Sandbox not available. Add sandbox tools to your agent config to enable it."
+            )
+            return
+
+        parsed = self.parse_args(args)
+        subcommand = parsed.positional.strip()
+
+        if subcommand.startswith("reset"):
+            # Parse session_id from after "reset"
+            rest = subcommand[len("reset"):].strip()
+            if parsed.has_flag("all"):
+                # Reset all sessions
+                sessions = manager.list_sessions()
+                if not sessions:
+                    app.session.add_message("system", "No active sandbox sessions.")
+                    return
+                for s in sessions:
+                    manager.reset_session(s["session_id"])
+                app.session.add_success(f"Reset {len(sessions)} sandbox session(s).")
+            else:
+                session_id = rest if rest else "default"
+                was_active = manager.reset_session(session_id)
+                if was_active:
+                    app.session.add_success(f"Sandbox session '{session_id}' reset.")
+                else:
+                    app.session.add_warning(f"Sandbox session '{session_id}' was not active.")
+        else:
+            # List sessions
+            sessions = manager.list_sessions()
+            if not sessions:
+                app.session.add_message("system", "No active sandbox sessions.")
+                return
+
+            table = Table(title="Sandbox Sessions", show_lines=False, padding=(0, 1))
+            table.add_column("Session ID", style="bold cyan", no_wrap=True)
+            table.add_column("Working Dir", style="dim")
+            table.add_column("Executions", style="dim", no_wrap=True, justify="right")
+
+            for s in sessions:
+                table.add_row(
+                    s["session_id"],
+                    s["working_dir"],
+                    str(s["execution_count"]),
+                )
+
+            app.session.add_rich(table)
+
+
 class PapersCommand(Command):
     """List documents in the knowledge base."""
 
@@ -227,3 +292,60 @@ class PapersCommand(Command):
             )
 
         app.session.add_rich(table)
+
+
+class SessionsCommand(Command):
+    """List and manage saved sessions."""
+
+    def __init__(self) -> None:
+        super().__init__(
+            name="sessions",
+            description="List saved sessions",
+            aliases=["sess"],
+            usage="/sessions [--delete=<id>]",
+            examples=["/sessions", "/sessions --delete=my-research"],
+            category=CommandCategory.SESSION,
+        )
+
+    async def execute(self, args: str, app: Any) -> None:
+        """Display saved sessions or delete one."""
+        from agentic_cli.persistence.session import SessionPersistence
+
+        parsed = self.parse_args(args)
+        delete_id = parsed.get_option("delete", "", str) or ""
+
+        persistence = SessionPersistence(app.settings)
+
+        if delete_id:
+            if persistence.delete_session(delete_id):
+                app.session.add_success(f"Session '{delete_id}' deleted.")
+            else:
+                app.session.add_error(f"Session '{delete_id}' not found.")
+            return
+
+        sessions = persistence.list_sessions()
+        if not sessions:
+            app.session.add_message("system", "No saved sessions.")
+            return
+
+        table = Table(title="Saved Sessions", show_lines=False, padding=(0, 1))
+        table.add_column("Session ID", style="bold cyan")
+        table.add_column("Messages", style="dim", justify="right")
+        table.add_column("Last Saved", style="dim")
+        table.add_column("Created", style="dim")
+
+        current_sid = app.session_id
+
+        for s in sessions:
+            sid = s["session_id"]
+            label = f"* {sid}" if sid == current_sid else sid
+            saved_at = s.get("saved_at", "")[:19].replace("T", " ")
+            created_at = s.get("created_at", "")[:10]
+            table.add_row(
+                label,
+                str(s.get("message_count", 0)),
+                saved_at,
+                created_at,
+            )
+
+        app.session.add_rich(table)

{agentic_cli-0.4.3 → agentic_cli-0.4.4}/src/agentic_cli/cli/commands.py

@@ -245,31 +245,6 @@ class Command(ABC):
 
         return [strip_quotes(token) for token in tokens]
 
-    def get_help(self) -> str:
-        """Get detailed help text for this command.
-
-        Returns:
-            Formatted help string
-        """
-        lines = [
-            f"**/{self.name}**",
-            f"  {self.description}",
-            "",
-            f"**Usage:** `{self.usage}`",
-        ]
-
-        if self.aliases:
-            lines.append(f"**Aliases:** {', '.join(f'/{a}' for a in self.aliases)}")
-
-        if self.examples:
-            lines.append("")
-            lines.append("**Examples:**")
-            for example in self.examples:
-                lines.append(f"  `{example}`")
-
-        return "\n".join(lines)
-
-
 class CommandRegistry:
     """Registry for managing slash commands.