deepagents-cli 0.0.30__tar.gz → 0.0.31__tar.gz

{deepagents_cli-0.0.30 → deepagents_cli-0.0.31}/CHANGELOG.md

@@ -1,5 +1,25 @@
 # Changelog
 
+## [0.0.31](https://github.com/langchain-ai/deepagents/compare/deepagents-cli==0.0.30...deepagents-cli==0.0.31) (2026-03-09)
+
+### Features
+
+* Opt-in `ask_user` tool for interactive agent questions ([#1377](https://github.com/langchain-ai/deepagents/issues/1377)) ([de7068d](https://github.com/langchain-ai/deepagents/commit/de7068d21fd4b932c6e53f500b0ea3b02a04c0aa))
+* Big thread improvements!
+  * Rework `/thread` switcher with search, columns, delete, and sort toggle ([#1723](https://github.com/langchain-ai/deepagents/issues/1723)) ([8b21ddb](https://github.com/langchain-ai/deepagents/commit/8b21ddb2ff7f13d6b3ffcbf2fe605bfbadbc3d38))
+  * Track and display working directory per thread ([#1735](https://github.com/langchain-ai/deepagents/issues/1735)) ([0e4f25d](https://github.com/langchain-ai/deepagents/commit/0e4f25dfbc3e15653bc3f8a6d32a0a61ead4ba82))
+  * Add `-n` short flag for `threads list --limit` ([#1731](https://github.com/langchain-ai/deepagents/issues/1731)) ([8bbace9](https://github.com/langchain-ai/deepagents/commit/8bbace9facd1e33757521e835dcb291accd2fa91))
+  * Add sort, branch filter, and verbose flags to threads list ([#1732](https://github.com/langchain-ai/deepagents/issues/1732)) ([11dc8e3](https://github.com/langchain-ai/deepagents/commit/11dc8e3397ef9e9dbe8b15578e9258544ed6b452))
+* Tailor system prompt for non-interactive mode ([#1727](https://github.com/langchain-ai/deepagents/issues/1727)) ([871e5cf](https://github.com/langchain-ai/deepagents/commit/871e5cf76b1a7e7cf7175b4415bb8e2206da39ec))
+* `/reload` command for in-session config refresh ([#1722](https://github.com/langchain-ai/deepagents/issues/1722)) ([381aee6](https://github.com/langchain-ai/deepagents/commit/381aee6d223fe3d866bedfe3a534916f419a4435))
+* Rearrange HITL option order in approval menu ([#1726](https://github.com/langchain-ai/deepagents/issues/1726)) ([0ca6cb2](https://github.com/langchain-ai/deepagents/commit/0ca6cb237b6da538bad2b4bf292942c8db72ec1f))
+
+### Bug Fixes
+
+* Localize newline shortcut labels by platform ([#1721](https://github.com/langchain-ai/deepagents/issues/1721)) ([f35576b](https://github.com/langchain-ai/deepagents/commit/f35576bafac711d6c04f1f9dd40ec97a90e30060))
+* Prevent `shift+enter` from sending `backslash+enter` ([#1728](https://github.com/langchain-ai/deepagents/issues/1728)) ([81dceb0](https://github.com/langchain-ai/deepagents/commit/81dceb043097a47702bb5a0227a8f12e9055bd05))
+* Write files with langsmith sandbox ([#1714](https://github.com/langchain-ai/deepagents/issues/1714)) ([5933c9e](https://github.com/langchain-ai/deepagents/commit/5933c9e2995c422e43649c61981e086ac1eaf725))
+
 ## [0.0.30](https://github.com/langchain-ai/deepagents/compare/deepagents-cli==0.0.29...deepagents-cli==0.0.30) (2026-03-07)
 
 ### Features

{deepagents_cli-0.0.30 → deepagents_cli-0.0.31}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: deepagents-cli
-Version: 0.0.30
+Version: 0.0.31
 Summary: Terminal interface for Deep Agents - interactive AI agent with file operations, shell access, and sub-agent capabilities.
 Project-URL: Homepage, https://docs.langchain.com/oss/python/deepagents/overview
 Project-URL: Documentation, https://reference.langchain.com/python/deepagents/
@@ -137,20 +137,25 @@ deepagents
 
 ## 🤔 What is this?
 
-Using an LLM to call tools in a loop is the simplest form of an agent. This architecture, however, can yield agents that are "shallow" and fail to plan and act over longer, more complex tasks.
+The fastest way to start using Deep Agents. `deepagents-cli` is a pre-built coding agent in your terminal — similar to Claude Code or Cursor — powered by any LLM that supports tool calling. One install command and you're up and running, no code required.
 
-Applications like "Deep Research", "Manus", and "Claude Code" have gotten around this limitation by implementing a combination of four things: a **planning tool**, **sub agents**, access to a **file system**, and a **detailed prompt**.
+**What the CLI adds on top of the SDK:**
 
-`deepagents` is a Python package that implements these in a general purpose way so that you can easily create a Deep Agent for your application. For a full overview and quickstart of Deep Agents, the best resource is our [docs](https://docs.langchain.com/oss/python/deepagents/overview).
-
-**Acknowledgements: This project was primarily inspired by Claude Code, and initially was largely an attempt to see what made Claude Code general purpose, and make it even more so.**
+- **Interactive TUI** — rich terminal interface with streaming responses
+- **Conversation resume** — pick up where you left off across sessions
+- **Web search** — ground responses in live information
+- **Remote sandboxes** — run code in isolated environments (Modal, Runloop, Daytona, & more)
+- **Persistent memory** — agent remembers context across conversations
+- **Custom skills** — extend the agent with your own slash commands
+- **Headless mode** — run non-interactively for scripting and CI
+- **Human-in-the-loop** — approve or reject tool calls before execution
 
 ## 📖 Resources
 
-- **[CLI Documentation](https://docs.langchain.com/oss/python/deepagents/cli/overview)** — Full documentation
-- **[CLI Source](https://github.com/langchain-ai/deepagents/tree/main/libs/cli)** — Full source code
-- **[Deep Agents SDK](https://github.com/langchain-ai/deepagents)** — The underlying agent harness
-- **[Chat LangChain](https://chat.langchain.com)** - Chat interactively with the docs
+- **[CLI Documentation](https://docs.langchain.com/oss/python/deepagents/cli/overview)**
+- **[Changelog](https://github.com/langchain-ai/deepagents/blob/main/libs/cli/CHANGELOG.md)**
+- **[Source code](https://github.com/langchain-ai/deepagents/tree/main/libs/cli)**
+- **[Deep Agents SDK](https://github.com/langchain-ai/deepagents)** — underlying agent harness
 
 ## 📕 Releases & Versioning
 
@@ -161,3 +166,7 @@ See our [Releases](https://docs.langchain.com/oss/python/release-policy) and [Ve
 As an open-source project in a rapidly developing field, we are extremely open to contributions, whether it be in the form of a new feature, improved infrastructure, or better documentation.
 
 For detailed information on how to contribute, see the [Contributing Guide](https://docs.langchain.com/oss/python/contributing/overview).
+
+## 🤝 Acknowledgements
+
+This project was primarily inspired by Claude Code, and initially was largely an attempt to see what made Claude Code general purpose, and make it even more so.

{deepagents_cli-0.0.30 → deepagents_cli-0.0.31}/README.md

@@ -35,20 +35,25 @@ deepagents
 
 ## 🤔 What is this?
 
-Using an LLM to call tools in a loop is the simplest form of an agent. This architecture, however, can yield agents that are "shallow" and fail to plan and act over longer, more complex tasks.
+The fastest way to start using Deep Agents. `deepagents-cli` is a pre-built coding agent in your terminal — similar to Claude Code or Cursor — powered by any LLM that supports tool calling. One install command and you're up and running, no code required.
 
-Applications like "Deep Research", "Manus", and "Claude Code" have gotten around this limitation by implementing a combination of four things: a **planning tool**, **sub agents**, access to a **file system**, and a **detailed prompt**.
+**What the CLI adds on top of the SDK:**
 
-`deepagents` is a Python package that implements these in a general purpose way so that you can easily create a Deep Agent for your application. For a full overview and quickstart of Deep Agents, the best resource is our [docs](https://docs.langchain.com/oss/python/deepagents/overview).
-
-**Acknowledgements: This project was primarily inspired by Claude Code, and initially was largely an attempt to see what made Claude Code general purpose, and make it even more so.**
+- **Interactive TUI** — rich terminal interface with streaming responses
+- **Conversation resume** — pick up where you left off across sessions
+- **Web search** — ground responses in live information
+- **Remote sandboxes** — run code in isolated environments (Modal, Runloop, Daytona, & more)
+- **Persistent memory** — agent remembers context across conversations
+- **Custom skills** — extend the agent with your own slash commands
+- **Headless mode** — run non-interactively for scripting and CI
+- **Human-in-the-loop** — approve or reject tool calls before execution
 
 ## 📖 Resources
 
-- **[CLI Documentation](https://docs.langchain.com/oss/python/deepagents/cli/overview)** — Full documentation
-- **[CLI Source](https://github.com/langchain-ai/deepagents/tree/main/libs/cli)** — Full source code
-- **[Deep Agents SDK](https://github.com/langchain-ai/deepagents)** — The underlying agent harness
-- **[Chat LangChain](https://chat.langchain.com)** - Chat interactively with the docs
+- **[CLI Documentation](https://docs.langchain.com/oss/python/deepagents/cli/overview)**
+- **[Changelog](https://github.com/langchain-ai/deepagents/blob/main/libs/cli/CHANGELOG.md)**
+- **[Source code](https://github.com/langchain-ai/deepagents/tree/main/libs/cli)**
+- **[Deep Agents SDK](https://github.com/langchain-ai/deepagents)** — underlying agent harness
 
 ## 📕 Releases & Versioning
 
@@ -59,3 +64,7 @@ See our [Releases](https://docs.langchain.com/oss/python/release-policy) and [Ve
 As an open-source project in a rapidly developing field, we are extremely open to contributions, whether it be in the form of a new feature, improved infrastructure, or better documentation.
 
 For detailed information on how to contribute, see the [Contributing Guide](https://docs.langchain.com/oss/python/contributing/overview).
+
+## 🤝 Acknowledgements
+
+This project was primarily inspired by Claude Code, and initially was largely an attempt to see what made Claude Code general purpose, and make it even more so.

deepagents_cli-0.0.31/deepagents_cli/_version.py

@@ -0,0 +1,3 @@
+"""Version information for `deepagents-cli`."""
+
+__version__ = "0.0.31"  # x-release-please-version

{deepagents_cli-0.0.30 → deepagents_cli-0.0.31}/deepagents_cli/agent.py

@@ -2,7 +2,9 @@
 
 from __future__ import annotations
 
+import logging
 import os
+import re
 import shutil
 import tempfile
 from pathlib import Path
@@ -50,6 +52,8 @@ from deepagents_cli.unicode_security import (
     summarize_issues,
 )
 
+logger = logging.getLogger(__name__)
+
 DEFAULT_AGENT_NAME = "agent"
 """The default agent name used when no `-a` flag is provided."""
 
@@ -136,12 +140,17 @@ def reset_agent(agent_name: str, source_agent: str | None = None) -> None:
     console.print(f"Location: {agent_dir}\n", style=COLORS["dim"])
 
 
-def get_system_prompt(assistant_id: str, sandbox_type: str | None = None) -> str:
+def get_system_prompt(
+    assistant_id: str,
+    sandbox_type: str | None = None,
+    *,
+    interactive: bool = True,
+) -> str:
     """Get the base system prompt for the agent.
 
-    Loads the immutable system prompt from `system_prompt.md` and
+    Loads the base system prompt template from `system_prompt.md` and
     interpolates dynamic sections (model identity, working directory,
-    skills path).
+    skills path, execution mode).
 
     Args:
         assistant_id: The agent identifier for path references
@@ -149,6 +158,8 @@ def get_system_prompt(assistant_id: str, sandbox_type: str | None = None) -> str
             (`'daytona'`, `'langsmith'`, `'modal'`, `'runloop'`).
 
             If `None`, agent is operating in local mode.
+        interactive: When `False`, the prompt is tailored for headless
+            non-interactive execution (no human in the loop).
 
     Returns:
         The system prompt string
@@ -166,6 +177,41 @@ def get_system_prompt(assistant_id: str, sandbox_type: str | None = None) -> str
 
     skills_path = f"~/.deepagents/{assistant_id}/skills/"
 
+    if interactive:
+        mode_description = "an interactive CLI on the user's computer"
+        interactive_preamble = (
+            "The user sends you messages and you respond with text and tool "
+            "calls. Your tools run on the user's machine. The user can see "
+            "your responses and tool outputs in real time, so keep them "
+            "informed — but don't over-explain."
+        )
+        ambiguity_guidance = (
+            "- If the request is ambiguous, ask questions before acting.\n"
+            "- If asked how to approach something, explain first, then act."
+        )
+    else:
+        mode_description = (
+            "non-interactive (headless) mode — there is no human operator "
+            "monitoring your output in real time"
+        )
+        interactive_preamble = (
+            "You received a single task and must complete it fully and "
+            "autonomously. There is no human available to answer follow-up "
+            "questions, so do NOT ask for clarification — make reasonable "
+            "assumptions and proceed."
+        )
+        ambiguity_guidance = (
+            "- Do NOT ask clarifying questions — there is no human to answer "
+            "them. Make reasonable assumptions and proceed.\n"
+            "- If you encounter ambiguity, choose the most reasonable "
+            "interpretation and note your assumption briefly.\n"
+            "- Always use non-interactive command variants — no human is "
+            "available to respond to prompts. Examples: `npm init -y` not "
+            "`npm init`, `apt-get install -y` not `apt-get install`, "
+            "`yes |` or `--no-input`/`--non-interactive` flags where "
+            "available. Never run commands that block waiting for stdin."
+        )
+
     # Build model identity section
     model_identity_section = ""
     if settings.model_name:
@@ -195,7 +241,14 @@ def get_system_prompt(assistant_id: str, sandbox_type: str | None = None) -> str
             f"- Use `{working_dir}` as your working directory for all operations\n\n"
         )
     else:
-        cwd = Path.cwd()
+        try:
+            cwd = Path.cwd()
+        except OSError:
+            logger.warning(
+                "Could not determine working directory for system prompt",
+                exc_info=True,
+            )
+            cwd = Path()
         working_dir_section = (
             f"### Current Working Directory\n\n"
             f"The filesystem backend is currently operating in: `{cwd}`\n\n"
@@ -208,12 +261,22 @@ def get_system_prompt(assistant_id: str, sandbox_type: str | None = None) -> str
             f"- Never use relative paths - always construct full absolute paths\n\n"
         )
 
-    return (
-        template.replace("{model_identity_section}", model_identity_section)
+    result = (
+        template.replace("{mode_description}", mode_description)
+        .replace("{interactive_preamble}", interactive_preamble)
+        .replace("{ambiguity_guidance}", ambiguity_guidance)
+        .replace("{model_identity_section}", model_identity_section)
         .replace("{working_dir_section}", working_dir_section)
         .replace("{skills_path}", skills_path)
     )
 
+    # Detect unreplaced placeholders (defense-in-depth for template typos)
+    unreplaced = re.findall(r"\{[a-z_]+\}", result)
+    if unreplaced:
+        logger.warning("System prompt contains unreplaced placeholders: %s", unreplaced)
+
+    return result
+
 
 def _format_write_file_description(
     tool_call: ToolCall, _state: AgentState[Any], _runtime: Runtime[Any]
@@ -431,10 +494,12 @@ def create_cli_agent(
     sandbox: SandboxBackendProtocol | None = None,
     sandbox_type: str | None = None,
     system_prompt: str | None = None,
+    interactive: bool = True,
     auto_approve: bool = False,
     enable_memory: bool = True,
     enable_skills: bool = True,
     enable_shell: bool = True,
+    enable_ask_user: bool = False,
     checkpointer: BaseCheckpointSaver | None = None,
     mcp_server_info: list[MCPServerInfo] | None = None,
 ) -> tuple[Pregel, CompositeBackend]:
@@ -456,7 +521,11 @@ def create_cli_agent(
             Used for system prompt generation.
         system_prompt: Override the default system prompt.
 
-            If `None`, generates one based on `sandbox_type` and `assistant_id`.
+            If `None`, generates one based on `sandbox_type`, `assistant_id`,
+            and `interactive`.
+        interactive: When `False`, the auto-generated system prompt is
+            tailored for headless non-interactive execution. Ignored when
+            `system_prompt` is provided explicitly.
         auto_approve: If `True`, no tools trigger human-in-the-loop
             interrupts — all calls (shell execution, file writes/edits,
             web search, URL fetch) run automatically.
@@ -467,6 +536,7 @@ def create_cli_agent(
         enable_skills: Enable `SkillsMiddleware` for custom agent skills
         enable_shell: Enable shell execution via `LocalShellBackend`
             (only in local mode). When enabled, the `execute` tool is available.
+        enable_ask_user: Enable the `ask_user` tool for interactive questioning.
         checkpointer: Optional checkpointer for session persistence.
 
             If `None`, uses `InMemorySaver` (no persistence across
@@ -523,6 +593,12 @@ def create_cli_agent(
     # Build middleware stack based on enabled features
     agent_middleware = []
 
+    # Add ask_user middleware (must be early so its tool is available)
+    if enable_ask_user:
+        from deepagents_cli.ask_user import AskUserMiddleware
+
+        agent_middleware.append(AskUserMiddleware())
+
     # Add memory middleware
     if enable_memory:
         memory_sources = [str(settings.get_user_agent_md_path(assistant_id))]
@@ -592,7 +668,9 @@ def create_cli_agent(
     # Get or use custom system prompt
     if system_prompt is None:
         system_prompt = get_system_prompt(
-            assistant_id=assistant_id, sandbox_type=sandbox_type
+            assistant_id=assistant_id,
+            sandbox_type=sandbox_type,
+            interactive=interactive,
         )
 
     # Configure interrupt_on based on auto_approve setting