Mark the tool call as successful. For `execute` calls that actually ran (a start time was recorded via `set_running`), the elapsed ru

Extract plain text content from a Static widget.

Format a token count into a human-readable short string. Args: count: Number of tokens. Returns: Formatted string like `'12.

Accumulate token counts for one completed LLM request. Updates both the session totals and the per-model breakdown. Args:

Look up the `HarnessProfile` for a model spec. Resolution order: 1. Exact match on `spec`. 2. Provider prefix (everything before the fir

Push the manager screen.

Build runtime configuration from environment variables. Args: env: Environment mapping to read. Defaults to `os.environ`.

Switch to a new model, preserving conversation history. This requires a server-backed interactive session. It sets a model override t

Check whether the terminal is in ASCII charset mode. Convenience wrapper so widgets can branch on charset without importing both `_detect_cha

Context manager to mock `get_config()` with a specific `thread_id`. Args: thread_id: The `thread_id` to return, or `None` to simulate mis

Read file content for the requested line range. Args: file_path: Absolute file path. offset: Line offset to start rea

Persist state, optionally updating agent_id / revision in the same call.

Create a StoreBackend backed by InMemoryStore, optionally pre-populated with files.

Construct an `OAuthClientProvider` for an MCP server. Args: server_name: MCP server name used in re-auth messages. server_url: Re

Edit a file by replacing string occurrences.

Start the agent runtime, scheduler, and channels.

Stop managed components and cancel in-flight agent work.

Create a minimal skill directory with a valid `SKILL.md`. Args: skill_dir: Directory to create the skill in (will be created if needed).

Read file content from the mapped real path.

Offload older messages to free context window space.

Create a fake chat model compatible with summarization middleware.

Mirror the current connection state onto the bottom status bar. The app-level welcome banner keeps rendering its regular footer while

Edit file content at the mapped real path.

Intercept and process large tool results before they're added to state. Args: tool_result: The tool result to potentially evict (

Run `install.sh` non-interactively with the fake tools on `PATH`. `start_new_session` detaches the controlling terminal so `/dev/tty` is unop

Load the forced or saved theme name, or return the default. Resolution order: 1. `DEEPAGENTS_CODE_THEME` env var (explicit override). If it

Build the LangGraph stream config dict. Stamps the shared `coding-agent-v1` trace-metadata contract (LSEN-277) via `build_coding_agent_metada

Persist `tokens` to disk, preserving any stored client info. Also records the absolute Unix-epoch expiry as a sidecar so a cold-start

r"""Validate and normalize file path for security. Ensures paths are safe to use by preventing directory traversal attacks and enforcing cons

Get a `SandboxProvider` instance for the specified provider (internal). Args: provider_name: Name of the provider. Resolved through the r

Check if a command contains dangerous shell patterns. These patterns can be used to bypass allow-list validation by embedding arbitrary comma

Build a client from `LANGSMITH_*` / `LANGCHAIN_*` env vars. Endpoint resolution is env var > `_DEFAULT_ENDPOINT`. Project-local deplo

Get available models dynamically from installed LangChain provider packages. Imports model profiles from each provider package and extracts model

Look up the `ProviderProfile` for a model spec. !!! beta `deepagents.profiles` exposes beta APIs that may receive minor changes in

Search files for literal text pattern. Routes to backends based on path: specific route searches one backend, `"/"` or `None` searche

Create a new agent session with the given working directory.

Write a JSON envelope to stdout and flush. The envelope is a single-line JSON object with a stable schema: ```json {"schema_version": 1,

Build the async subagent tools from agent specs. Args: agents: List of async subagent specifications. Returns: List of `Stru

Format tool output based on tool type for nicer display. Args: output: Raw output string is_preview: Whether this is

Build a mock MCP `Tool` object suitable for conversion.

Create a mock ToolRuntime with a real dict for state.

Handle escape key. Priority order: 1. If modal screen is active, dismiss it 2. If completion popup is open, dismiss it

Async version of `ls`, delegating to `aexecute`.

Cancel the current execution.

Write content to a file, creating it or overwriting it if it already exists. The update is queued directly via `CONFIG_KEY_SEND`.

Create a `WelcomeBanner` with all env vars cleared. Args: thread_id: Optional thread ID to display. project_name: If set, simulat

Build a mock tool as the adapter produces with `tool_name_prefix=True`.

Register the LangChain theme on a test app so ThemeSelectorScreen works.

Store the client connection for sending session updates.

Show the model selector screen. Starts in the full-list (`_recommended_only=False`) state so that legacy assertions about the full ca

A synchronous tool that records its invocations. Async-capable by default because `StructuredTool.from_function` synthesises a coroutine wrap

Handle Ctrl+C - interrupt agent, reject approval, or quit on double press. Priority order: 1. If a focused input has a non-empty sele

Get all tracked images. Returns: Copy of the list of tracked images.

Return credential readiness details for a provider. Combines config, well-known provider metadata, optional provider auth, and implicit-auth

Return whether env var *name* is set to a recognizably truthy value. Unlike `bool(os.environ.get(name))`, this does not treat `"0"` or `"fals

Create SKILL.md content with YAML frontmatter. Args: name: Skill name for frontmatter description: Skill description for frontmat

Set the future to resolve when user answers.

Execute the side effect for a notification action. Catches `Exception` broadly so any failure in the handler surfaces as a warning to

Create a mock backend with execute() returning the given output.

Apply enabled auto-updates before the TUI and server start. On a successful upgrade the process is re-exec'd so the new version is loaded. An

Search for a literal text pattern in files. Uses ripgrep if available, falling back to Python search. Args: pattern: Lit

List directory contents (non-recursive). If path matches a route, lists only that backend. If path is `"/"`, aggregates default backe

Resolve a model string to a `BaseChatModel`. If `model` is already a `BaseChatModel`, returns it unchanged. String models are resolved via `

Create a mock BaseChatModel exposing `attrs` via attribute access. `get_model_identifier` reads `model_name` / `model` directly off the insta

Build a mock `list_tools` page result.

Return a minimal AgentMiddleware whose .name returns `name`.

Build a single-`ToolCallMessage` Textual app for pilot-driven tests. Args: tool_name: Tool name the message represents. args: Opt

Async version of grep. See `grep()` for detailed documentation on routing behavior and parameters.

Run the `deepagents mcp-servers` sub-command.

Get the base system prompt for the agent. Loads the base system prompt template from `system_prompt.md` and interpolates dynamic sections (mo

Clear all phases and hide the panel immediately (e.g. on `/clear`).

Run agent eval asynchronously against the given query. Args: agent: The compiled state graph to invoke. query: A string prompt or

Start the `langgraph dev` server and wait for it to be healthy. Args: timeout: Max seconds to wait for the server to become healt

Load and render message history when resuming a thread. When `preloaded_payload` is provided (e.g., from `_resume_thread`), this reus

Create a RemoteAgent pointing at a dummy URL for test scaffolding.

Parse a PEP 440 version string into a comparable `Version` object. Supports stable (`1.2.3`) and pre-release (`1.2.3a1`, `1.2.3rc2`) versions.